Kobweb is an opinionated Kotlin framework for creating websites and web apps, built on top of Compose HTML and inspired by Next.js and Chakra UI.
@Page
@Composable
fun HomePage() {
Column(Modifier.fillMaxWidth(), horizontalAlignment = Alignment.CenterHorizontally) {
Row(Modifier.align(Alignment.End)) {
var colorMode by ColorMode.currentState
Button(
onClick = { colorMode = colorMode.opposite },
Modifier.borderRadius(50.percent).padding(0.px)
) {
// Includes support for Font Awesome icons
if (colorMode.isLight) FaSun() else FaMoon()
}
}
H1 {
Text("Welcome to Kobweb!")
}
Row(Modifier.flexWrap(FlexWrap.Wrap)) {
SpanText("Create rich, dynamic web apps with ease, leveraging ")
Link("https://kotlinlang.org/", "Kotlin")
SpanText(" and ")
Link("https://github.com/JetBrains/compose-multiplatform#compose-html/", "Compose HTML")
}
}
}
While Kobweb is still pre-1.0, it has been usable for a while now. It provides escape hatches to lower-level APIs, so you can accomplish anything even if Kobweb doesn't support it yet. Please consider starring the project to indicate interest, so we know we're creating something the community wants. How ready is it?▼
Our goal is to provide:
- an intuitive structure for organizing your Kotlin website or web app
- automatic handling of routing between pages
- a collection of useful batteries included widgets built on top of Compose HTML
- an environment built from the ground up around live reloading
- static site exports for improved SEO
- support for responsive (i.e. mobile and desktop) design
- out-of-the-box Markdown support
- a way to easily define server API routes and persistent API streams
- support for creating and using web workers
- a growing collection of general purpose utilities added on top of Compose HTML and Kotlin/JS (learn more▼)
- an open source foundation that the community can extend
- and much, much more!
Here's a demo where we create a Compose HTML project from scratch with Markdown support and live reloading, in under 10 seconds:
kobweb-demo.mp4
You can check out my talk at Droidcon SF 24 for a high level overview of Kobweb. The talk showcases what Kobweb can do, introduces Compose HTML (which it builds on top of), and covers a wide range of frontend and backend functionality. It is light on code but heavy on understanding the structure and capabilities of the framework.
One of Kobweb's users, Stevdza-San, has created free starting tutorials that demonstrate how to build projects using Kobweb.
- Getting started with Kobweb
- This video introduces basic Kobweb concepts and walks you through the process from creating a simple (static layout) site to exporting it locally on your machine (with files you can then upload to a static hosting provider of your choice).
- Deploying a Kobweb site
- This video builds upon the previous, showcasing some additional tips and tricks, and walks you all the way through to deploying your site live on the internet using free hosting.
- Building a full stack multiplatform site
- This demonstrates how to write both frontend and backend logic. It also demonstrates how you can write a separate Android frontend that can also work with your server. (This video is still useful to watch even if you never intend to use Android).
Tip
It's easy to start with a static layout site first and migrate to a full stack site later. (You can read more about Static layout vs. Full stack sites▼ below.)
A YouTube channel called SkyFish created a tutorial video on how to build a fullstack site with Kobweb.
The first step is to get the Kobweb binary. You can install it, download it, and/or build it, so we'll include instructions for all these approaches.
Major thanks to aalmiray and helpermethod to helping me get these installation options working. Check out JReleaser if you ever need to do this in your own project!
OS: Mac and Linux
$ brew install varabyte/tap/kobweb
OS: Windows
# Note: Adding buckets only has to be done once.
# Feel free to skip java if you already have it
> scoop bucket add java
> scoop install java/openjdk
# Install kobweb
> scoop bucket add varabyte https://github.com/varabyte/scoop-varabyte.git
> scoop install varabyte/kobweb
OS: Windows, Mac, and *nix
$ sdk install kobweb
Thanks a ton to aksh1618 for adding support for this target!
With an AUR helper, e.g.:
$ yay -S kobweb
$ paru -S kobweb
$ trizen -S kobweb
# etc.
Without an AUR helper:
$ git clone https://aur.archlinux.org/kobweb.git
$ cd kobweb
$ makepkg -si
Please see: varabyte/kobweb-cli#11 and consider leaving a comment!
Our binary artifact is hosted on GitHub. To download the latest, you can either grab the zip or tar file from GitHub or you can fetch it from your terminal:
$ cd /path/to/applications
# You can either pull down the zip file
$ wget https://github.com/varabyte/kobweb-cli/releases/download/v0.9.18/kobweb-0.9.18.zip
$ unzip kobweb-0.9.18.zip
# ... or the tar file
$ wget https://github.com/varabyte/kobweb-cli/releases/download/v0.9.18/kobweb-0.9.18.tar
$ tar -xvf kobweb-0.9.18.tar
and I recommend adding it to your path, either directly:
$ PATH=$PATH:/path/to/applications/kobweb-0.9.18/bin
$ kobweb version # to check it's working
or via symbolic link:
$ cd /path/to/bin # some folder you've created that's in your PATH
$ ln -s /path/to/applications/kobweb-0.9.18/bin/kobweb kobweb
Although we host Kobweb artifacts on GitHub, it's easy enough to build your own.
Building Kobweb requires JDK11 or newer. We'll first discuss how to add it.
If you want full control over your JDK install, manually downloading is a good option.
- Download a JDK for your OS
- Unzip it somewhere
- Update your
JAVA_HOME
variable to point at it.
JAVA_HOME=/path/to/jdks/corretto-11.0.12
# ... or whatever version or path you chose
For a more automated approach, you can request IntelliJ install a JDK for you.
Follow their instructions here: https://www.jetbrains.com/help/idea/sdk.html#set-up-jdk
The Kobweb CLI is actually maintained in a separate GitHub repo. Once you have the JDK set up, it should be easy to clone and build it:
$ cd /path/to/src/root # some folder you've created for storing src code
$ git clone https://github.com/varabyte/kobweb-cli
$ cd kobweb-cli
$ ./gradlew :kobweb:installDist
Finally, update your PATH:
$ PATH=$PATH:/path/to/src/root/kobweb-cli/kobweb/build/install/kobweb/bin
$ kobweb version # to check it's working
If you previously installed Kobweb and are aware that a new version is available, the way you update it depends on how you installed it.
Method | Instructions |
---|---|
Homebrew | brew update brew upgrade kobweb |
Scoop | scoop update kobweb |
SDKMAN! | sdk upgrade kobweb |
Arch Linux | Rerunning install steps should work. If using an AUR helper, you may need to review its manual. |
Downloaded from Github |
Visit the latest release. You can find both a zip and tar file there. |
$ cd /path/to/projects/
$ kobweb create app
You'll be asked a few questions required for setting up your project.
You don't need to create a root folder for your project ahead of time - the setup process will prompt you for one to create. For the remaining parts of this section, let's say you choose the folder "my-project" when asked.
When finished, you'll have a basic project with two pages - a home page and an about page (with the about page written in markdown) - and some components (which are collections of reusable, composable pieces). Your own directory structure should look something like this:
my-project
└── site/src/jsMain
├── kotlin.org.example.myproject
│ ├── components
│ │ ├── layouts
│ │ │ ├── MarkdownLayout.kt
│ │ │ └── PageLayout.kt
│ │ ├── sections
│ │ │ ├── Footer.kt
│ │ │ └── NavHeader.kt
│ │ └── widgets
│ │ └── IconButton.kt
│ ├── pages
│ │ └── Index.kt
│ └── AppEntry.kt
└── resources/markdown
└── About.md
Note that there's no index.html or routing logic anywhere! We generate that for you automatically when you run Kobweb. This brings us to the next section...
$ cd your-project/site
$ kobweb run
This command spins up a web server at http://localhost:8080. If you want to configure the port, you can do so by editing
your project's .kobweb/conf.yaml
file.
You can open your project in IntelliJ and start editing it. While Kobweb is running, it will detect changes, recompile, and deploy updates to your site automatically.
If you don't want to keep a separate terminal window open beside your IDE window, you may prefer alternate solutions.
You can use the IntelliJ terminal tool window to run
kobweb
within it. If you run into a compile error, the stack trace lines will get decorated with
links, making it easy to navigate to the relevant source.
kobweb
itself delegates to Gradle, but nothing is stopping you from calling the commands yourself. You can create
Gradle run configurations for each of the Kobweb commands.
Tip
When you run a Kobweb CLI command that delegates to Gradle, it will log the Gradle command to the console. This is how you can discover the Gradle commands discussed in this section.
- To start a Kobweb server, use the
kobwebStart -t
command.- The
-t
argument (or,--continuous
) tells Gradle to watch for file changes, which gives you live loading behavior.
- The
- To stop a running Kobweb server, use the
kobwebStop
command. - To export a site, use
kobwebExport -PkobwebReuseServer=false -PkobwebEnv=DEV -PkobwebRunLayout=FULLSTACK -PkobwebBuildTarget=RELEASE -PkobwebExportLayout=FULLSTACK
- If you want to export a static layout instead, change the last argument to
-PkobwebExportLayout=STATIC
.
- If you want to export a static layout instead, change the last argument to
- To run an exported site, use
kobwebStart -PkobwebEnv=PROD -PkobwebRunLayout=FULLSTACK
- If your site was exported using a static layout, change the last argument to
-PkobwebRunLayout=STATIC
.
- If your site was exported using a static layout, change the last argument to
You can read all about IntelliJ's Gradle integration here. Or to just jump straight into how to create run configurations for any of the commands discussed above, read these instructions.
Kobweb will provide a growing collection of samples for you to learn from. To see what's available, run:
$ kobweb list
You can create the following Kobweb projects by typing `kobweb create ...`
• app: A template for a minimal site that demonstrates the basic features of Kobweb
• examples/jb/counter: A very minimal site with just a counter (based on the Jetbrains tutorial)
• examples/todo: An example TODO app, showcasing client / server interactions
For example, kobweb create examples/todo
will instantiate a TODO app locally.
The project templates created by Kobweb all embrace Gradle version catalogs.
If you're not aware of it, it's a file that exists at gradle/libs.versions.toml
. If you find yourself wanting to tweak
or add new versions to projects you originally created via kobweb create
, that's where you'll find them.
For example, here's the libs.versions.toml we use for our own landing site.
To read more about the feature, please check out the official docs.
The latest available version of Kobweb is declared at the top of this README. If a new version has come out, you can
update your own project by editing gradle/libs.version.toml
and updating the kobweb
version there.
Important
You should double-check COMPATIBILITY.md to see if you also need to update your kotlin
and
jetbrains-compose
versions as well.
Caution
It can be confusing, but Kobweb has two versions -- the version for the library itself (the one that is applicable in this situation), and the one for the command line tool.
Kobweb, at its core, is a handful of classes responsible for trimming away much of the boilerplate around building a Compose HTML app, such as routing and configuring basic CSS styles. Kobweb further provides a Gradle plugin which analyzes your codebase and generates relevant boilerplate code.
Kobweb is also a CLI binary of the same name which provides commands to handle the tedious parts of building and/or running a Compose HTML app. We want to get that stuff out of the way, so you can enjoy focusing on the more interesting work!
(To learn more about Compose HTML, please visit the official tutorials).
Creating a page is easy! It's just a normal @Composable
method. To upgrade your composable to a page, all you need to
do is:
- Define your composable in a file somewhere under the
pages
package in yourjsMain
source directory. - Annotate it with
@Page
Just from that, Kobweb will create a site entry for you automatically.
For example, if I create the following file:
// jsMain/kotlin/com/mysite/pages/admin/Settings.kt
@Page
@Composable
fun SettingsPage() {
/* ... */
}
this will create a page that I can then visit by going to mysite.com/admin/settings
.
Important
The last part of a URL, here settings
, is called a slug.
By default, the slug comes from the file name, which is converted into kebab-case, e.g. AboutUs.kt
would transform into
about-us
. However, this can be overridden to whatever you want (more on that shortly).
The file name Index.kt
is special. If a page is defined inside such a file, it will be treated as the default page
under that URL. For example, a page defined in .../pages/admin/Index.kt
will be visited if the user visits
mysite.com/admin/
.
If you ever need to change the route generated for a page, you can set the Page
annotation's routeOverride
field:
// jsMain/kotlin/com/mysite/pages/admin/Settings.kt
@Page(routeOverride = "config")
@Composable
fun SettingsPage() {
/* ... */
}
The above would create a page you could visit by going to mysite.com/admin/config
.
routeOverride
can additionally contain slashes, and if the value begins and/or ends with a slash, that has a special
meaning.
- Begins with a slash - represent the whole route from the root
- Ends with a slash - a slug will still be generated from the filename and appended to the route override.
And if you set the override to "index", that behaves the same as setting the file to Index.kt
as described above.
Some examples can clarify these rules (and how they behave when combined). Assuming we're defining a page for our site
example.com
within the file a/b/c/Slug.kt
:
Annotation | Resulting URL |
---|---|
@Page |
example.com/a/b/c/slug |
@Page("other") |
example.com/a/b/c/other |
@Page("index") |
example.com/a/b/c/ |
@Page("d/e/f/") |
example.com/a/b/c/d/e/f/slug |
@Page("d/e/f/other") |
example.com/a/b/c/d/e/f/other |
@Page("d/e/f/index") |
example.com/a/b/c/d/e/f/ |
@Page("/d/e/f/") |
example.com/d/e/f/slug |
@Page("/d/e/f/other") |
example.com/d/e/f/other |
@Page("/d/e/f/index") |
example.com/d/e/f/ |
@Page("/") |
example.com/slug |
@Page("/other") |
example.com/other |
@Page("/index") |
example.com/ |
Caution
Despite the flexibility allowed here, you should not be using this feature frequently, if at all. A Kobweb project benefits from the fact that a user can easily associate a URL on your site with a file in your codebase, but this feature allows you to break those assumptions. It is mainly provided to enable dynamic routing (see the Dynamic Routes▼ section) or enabling a URL name that uses characters which aren't allowed in Kotlin filenames.
While the slug is derived from the filename, earlier parts of the route are derived from the file's package.
A package will be converted into a route part by removing any leading or trailing underscores (as these are often used
to work around limitations into what values and keywords are allowed in a package name, e.g. site.pages.blog._2022
and
site.events.fun_
) and converting camelCase packages into hyphenated words (so site.pages.team.ourValues
generates
the route /team/our-values/
).
If you'd like to override the route part generated for a package, you can use the PackageMapping
annotation.
For example, let's say your team prefers not to use camelCase packages for aesthetic reasons. Or perhaps you
intentionally want to add a leading underscore into your site's route part for some emphasis (since earlier we mentioned
that leading underscores get removed automatically), such as in the route /team/_internal/contact-numbers
. You can
use package mappings for this.
You apply the package mapping annotation to the current file. Using it looks like this:
// site/pages/team/values/PackageMapping.kt
@file:PackageMapping("our-values")
package site.pages.blog.values
import com.varabyte.kobweb.core.PackageMapping
With the above package mapping in place, a file that lives at site/pages/team/values/Mission.kt
will be visitable at
/team/our-values/mission
.
Every page method provides access to its PageContext
via the rememberPageContext()
method.
Critically, a page's context provides it access to a router, allowing you to navigate to other pages.
It also provides dynamic information about the current page's URL (discussed in the next section).
@Page
@Composable
fun ExamplePage() {
val ctx = rememberPageContext()
Button(onClick = { ctx.router.navigateTo("/other/page") }) {
Text("Click me")
}
}
You can use the page context to check the values of any query parameters passed into the current page's URL.
So if you visit site.com/posts?id=12345&mode=edit
, you can query those values like so:
enum class Mode {
EDIT, VIEW;
companion object {
fun from(value: String) {
entries.find { it.name.equals(value, ignoreCase = true) }
?: error("Unknown mode: $value")
}
}
}
@Page
@Composable
fun Posts() {
val ctx = rememberPageContext()
// Here, I'm assuming these params are always present, but you can use
// `get` instead of `getValue` to handle the nullable case. Care should
// also be taken to parse invalid values without throwing an exception.
val postId = ctx.route.params.getValue("id").toInt()
val mode = Mode.from(ctx.route.params.getValue("mode"))
/* ... */
}
In addition to query parameters, Kobweb supports embedding arguments directly in the URL itself. For example, you might
want to register the path users/{user}/posts/{post}
which would be visited if the site visitor typed in a URL like
users/bitspittle/posts/20211231103156
.
How do we set it up? Thankfully, it's fairly easy.
But first, notice that in the example dynamic route users/{user}/posts/{post}
there are actually two different dynamic
parts, one in the middle and one at the tail end. These can be handled by the PackageMapping
and Page
annotations,
respectively.
Pay attention to the use of the curly braces in the mapping name! That lets Kobweb know that this is a dynamic package.
// pages/users/user/PackageMapping.kt
@file:PackageMapping("{user}") // or @file:PackageMapping("{}")
package site.pages.users.user
import com.varabyte.kobweb.core.PackageMapping
If you pass an empty "{}"
into the PackageMapping
annotation, it directs Kobweb to use the name of the package
itself (i.e. user
in this specific case).
Like PackageMapping
, the Page
annotation can also take curly braces to indicate a dynamic value.
// pages/users/user/posts/Post.kt
@Page("{post}") // Or @Page("{}")
@Composable
fun PostPage() {
/* ... */
}
An empty "{}"
tells Kobweb to use the name of the current file.
Remember that the Page
annotation allows you to rewrite the entire route. That value also accepts dynamic parts, so
you could even do something like:
// pages/users/user/posts/Post.kt
@Page("/users/{user}/posts/{post}") // Or @Page("/users/{user}/posts/{}")
@Composable
fun PostPage() {
/* ... */
}
but with great power comes great responsibility. Tricks like this may be hard to find and/or update later, especially as your project gets larger. While it works, you should only use this format in cases where you absolutely need to (perhaps after a code refactor where you have to support legacy URL paths).
You query dynamic route values exactly the same as if you were requesting query parameters. That is, use ctx.params
:
@Page("{}")
@Composable
fun PostPage() {
val ctx = rememberPageContext()
val postId = ctx.route.params.getValue("post")
/* ... */
}
Important
You should avoid creating URL paths where the dynamic path and the query parameters have the same name, as in
mysite.com/posts/{post}?post=...
, as this could be really tricky to debug in a complex project. If there is a
conflict, then the dynamic route parameters will take precedence. (You can still access the query parameter value via
ctx.route.queryParams
in this case if necessary.)
If you have a resource that you'd like to serve from your site, you handle this by placing it in your site's
jsMain/resources/public
folder.
For example, if you have a logo you'd like to be available at mysite.com/assets/images/logo.png
, you would put it in
your Kobweb project at jsMain/resources/public/assets/images/logo.png
.
In other words, anything under your project resources' public/
directory will be automatically copied over to your
final site (not including the public/
part).
For those new to web dev, it's worth understanding that there are two ways to set styles on your HTML elements: inline and stylesheet.
Inline styles are defined on the element tag itself. In raw HTML, this might look like:
<div style="background-color:black">
Meanwhile, any given HTML page can reference a list of stylesheets which can define a bunch of styles, where each style is tied to a selector (a rule which selects what elements those styles apply to).
A concrete example of a very short stylesheet can help here:
body {
background-color: black;
color: magenta
}
#title {
color: yellow
}
And you could use that stylesheet to style the following document:
<body>
<!-- Title gets background-color from "body" and foreground color from "#title" -->
<div id="title">Yellow on black</div>
Magenta on black
</body>
Note
When conflicting styles are present both in a stylesheet and as an inline declaration, the inline styles take precedence.
There's no hard and fast rule, but in general, when writing HTML / CSS by hand, stylesheets are often preferred over inline styles as it better maintains a separation of concerns. That is, the HTML should represent the content of your site, while the CSS controls the look and feel.
However! We're not writing HTML / CSS by hand. We're using Compose HTML! Should we even care about this in Kotlin?
As it turns out, there are times when you have to use stylesheets, because without them, you can't define styles for advanced behaviors (particularly pseudo-classes, pseudo-elements, and media queries). For example, you can't override the color of visited links without using a stylesheet approach. So it's worth realizing there are fundamental differences.
Finally, it can also be much easier debugging your page with browser tools when you lean on stylesheets over inline styles, as it
makes your DOM tree easier to read when your elements are simple (e.g. <div class="title">
vs. <div style="color:yellow; background-color:black; font-size: 24px; ...">
).
We'll be introducing and discussing modifiers and CSS style blocks in more detail shortly. But in general, when you pass modifiers directly into a composable widget in Silk, those will result in inline styles, whereas if you use a CSS style block to define your styles, those will get embedded into the site's stylesheet:
// Uses inline styles
Box(Modifier.color(Colors.Red)) { /* ... */ }
// Uses a stylesheet
val BoxStyle = CssStyle {
base { Modifier.Color(Colors.Red) }
}
Box(BoxStyle.toModifier()) { /* ... */ }
As a beginner, or even as an advanced user when prototyping, feel free to use inline modifiers as much as you can, pivoting to CSS style blocks if you find yourself needing to use pseudo-classes, pseudo-elements, or media queries. It is fairly easy to migrate inline styles over to stylesheets in Kobweb.
In my own projects, I tend to use inline styles for really simple layout elements (e.g. Row(Modifier.fillMaxWidth())
)
and CSS style blocks for complex and/or re-usable widgets. It actually becomes a nice organizational convention to have
all your styles grouped together in one place above the widget itself.
Kobweb introduces the Modifier
class, in order to provide an experience similar to what you find in Jetpack Compose.
(You can read more about them here if you're unfamiliar with
the concept).
In the world of Compose HTML, you can think of a Modifier
as a wrapper on top of CSS styles and attributes.
Important
Please refer to official documentation if you are not familiar with HTML attributes and/or styles.
So this:
Modifier.backgroundColor(Colors.Red).color(Colors.Green).padding(200.px)
when passed into a widget provided by Kobweb, like Box
:
Box(Modifier.backgroundColor(Colors.Red).color(Colors.Green).padding(200.px)) {
/* ... */
}
would generate an HTML tag with a style property like: <div style="background:red;color:green;padding:200px">
There are a bunch of modifier extensions (and they're growing) provided by Kobweb, like background
, color
, and
padding
above. But there are also two escape hatches anytime you run into a modifier that's missing:
attrsModifier
and styleModifier
.
At this point, you are interacting with Compose HTML, one layer underneath Kobweb.
Using them looks like this:
// Modify attributes of an element tag
// e.g. the "a", "b", and "c" in <tag a="..." b="..." c="..." />
Modifier.attrsModifier {
id("example")
}
// Modify styles of an element tag
// e.g. the "x", "y", and "z" in `<tag a="..." b="..." c="..." style="x:...;y:...;z:..." />
Modifier.styleModifier {
width(100.percent)
height(50.percent)
}
// Note: Because "style" itself is an attribute, you can define styles in an attrsModifier:
Modifier.attrsModifier {
id("example")
style {
width(100.percent)
height(50.percent)
}
}
// ... but in the above case, you should use a styleModifier for simplicity
In the occasional (and hopefully rare!) case where Kobweb doesn't provide a modifier and Compose HTML doesn't provide
the attribute or style support you need, you can use attrsModifier
plus the attr
method or styleModifier
plus the
property
method. This escape hatch within an escape hatch allows you to provide any custom value you need.
The above cases can be rewritten as:
Modifier.attrsModifier {
attr("id", "example")
}
Modifier.styleModifier {
property("width", 100.percent)
// Or even raw CSS:
// property("width", "100%")
property("height", 50.percent)
}
If you end up needing to use attr
or property
in your own codebase, consider
filing an issue
with us so that we can add the missing modifier to the library.
Silk is a UI layer included with Kobweb and built upon Compose HTML.
While Compose HTML requires you to understand underlying HTML / CSS concepts, Silk attempts to abstract some of that away, providing an API more akin to what you might experience developing a Compose app on Android or Desktop. Less "div, span, flexbox, attrs, styles, classes" and more "Rows, Columns, Boxes, and Modifiers".
We consider Silk a pretty important part of the Kobweb experience, but it's worth pointing out that it's designed as an optional component. You can absolutely use Kobweb without Silk. (You can also use Silk without Kobweb!).
You can also interleave Silk and Compose HTML components easily (as Silk is just composing them itself).
Before going further, we want to quickly mention you can annotate a method with @InitSilk
, which will be called when
your site starts up.
This method must take a single InitSilkContext
parameter. A context contains various properties that allow
adjusting Silk defaults, which will be demonstrated in more detail in sections below.
@InitSilk
fun initSilk(ctx: InitSilkContext) {
// `ctx` has a handful of properties which allow you to adjust Silk's default behavior.
}
Tip
The names of your @InitSilk
methods don't matter, as long as they're public, take a single InitSilkContext
parameter, and don't collide with another method of the same name. You are encouraged to choose a name for readability
purposes.
You can define as many @InitSilk
methods as you want, so feel free to break them up into relevant, clearly named
pieces, instead of declaring a single, monolithic, generically named fun initSilk(ctx)
method that does everything.
With Silk, you can define a style like so, using the CssStyle
function and the base
block:
val CustomStyle = CssStyle {
base {
Modifier.background(Colors.Red)
}
}
and convert it to a modifier by using CustomStyle.toModifier()
. At this point, you can pass it into any composable
which takes a Modifier
parameter:
// Approach #1 (uses inline styles)
Box(Modifier.backgroundColor(Colors.Red)) { /* ... */ }
// Approach #2 (uses stylesheets)
Box(CustomStyle.toModifier()) { /* ... */ }
Important
When you declare a CssStyle
, it must be public. This is because code gets generated inside a main.kt
file by
the Kobweb Gradle plugin, and that code needs to be able to access your style in order to register it.
In general, it's a good idea to think of styles as global anyway, since technically they all live in a globally applied stylesheet, and you have to make sure that the style name is unique across your whole application.
You can technically make a style private if you add a bit of boilerplate to handle the registration yourself:
@Suppress("PRIVATE_COMPONENT_STYLE")
private val ExampleCustomStyle = CssStyle { /* ... */ }
// Or use a leading underscore to automatically suppress the warning
private val _ExampleOtherCustomStyle = CssStyle { /* ... */ }
@InitSilk
fun registerPrivateStyle(ctx: InitSilkContext) {
// Kobweb will not be able to detect the property name, so a name must be provided manually
ctx.theme.registerStyle("example-custom", ExampleCustomStyle)
ctx.theme.registerStyle("example-other-custom", _ExampleOtherCustomStyle)
}
However, you are encouraged to keep your styles public and let the Kobweb Gradle plugin handle everything for you.
You can simplify the syntax of basic style blocks a bit further with the CssStyle.base
declaration:
val CustomStyle = CssStyle.base {
Modifier.background(Colors.Red)
}
Just be aware you may have to break this out again if you find yourself needing to support additional selectors▼.
The Kobweb Gradle plugin automatically detects your CssStyle
properties and generates a name for it for you, derived
from the property name itself but
using Kebab Case.
For example, if you write val TitleTextStyle = CssStyle { ... }
, its name will be "title-text".
You usually won't need to care about this name, but there are niche cases where it can be useful to understand that is what's going on.
If you need to set a name manually, you can use the CssName
annotation to override the default name:
@CssName("my-custom-name")
val CustomStyle = CssStyle {
base {
Modifier.background(Colors.Red)
}
}
So, what's up with the base
block?
True, it looks a bit verbose on its own. However, you can define additional selector blocks that take effect conditionally. The base style will always apply first, but then any additional styles will be applied based on the specific selector's rules.
Caution
Order matters when defining additional selectors, particularly if more than one of them modify the same CSS property at the same time.
Here, we create a style which is red by default but green when the mouse hovers over it:
val CustomStyle = CssStyle {
base {
Modifier.color(Colors.Red)
}
hover {
Modifier.color(Colors.Green)
}
}
Kobweb provides a bunch of standard selectors for you for convenience, but for those who are CSS-savvy, you can always define the CSS rule directly to enable more complex combinations or selectors that Kobweb hasn't added yet.
For example, this is identical to the above style definition:
val CustomStyle = CssStyle {
base {
Modifier.color(Colors.Red)
}
cssRule(":hover") {
Modifier.color(Colors.Green)
}
}
There's a feature in the world of responsive HTML / CSS design called breakpoints, which confusingly have nothing to do with debugging breakpoints. Rather, they specify size boundaries for your site when styles change. This is how sites present content differently on mobile vs. tablet vs. desktop.
Kobweb provides four breakpoint sizes you can use for your project, which, including using no breakpoint size at all, gives you five buckets you can work with when designing your site:
- no breakpoint - mobile (and larger)
- sm - tablets (and larger)
- md - desktops (and larger)
- lg - widescreen (and larger)
- xl - ultra widescreen (and larger)
You can change the default values of breakpoints for your site by adding
an @InitSilk
method to your code and setting ctx.theme.breakpoints
:
@InitSilk
fun initializeBreakpoints(ctx: InitSilkContext) {
ctx.theme.breakpoints = BreakpointSizes(
sm = 30.cssRem,
md = 48.cssRem,
lg = 62.cssRem,
xl = 80.cssRem,
)
}
To reference a breakpoint in a CssStyle
, just invoke it:
val CustomStyle = CssStyle {
base {
Modifier.fontSize(24.px)
}
Breakpoint.MD {
Modifier.fontSize(32.px)
}
}
Tip
When testing your breakpoint-conditional styles, you should be aware that browser dev tools let you simulate window dimensions to see how your site looks at different sizes. For example, on Chrome, you can follow these instructions: https://developer.chrome.com/docs/devtools/device-mode
When you define a CssStyle
, a field called colorMode
is available for you to use:
val CustomStyle = CssStyle.base {
Modifier.color(if (colorMode.isLight) Colors.Red else Colors.Pink)
}
Silk defines a bunch of light and dark colors for all of its widgets, and if you'd like to re-use any of them in your
own widget, you can query them using colorMode.toPalette()
:
val CustomStyle = CssStyle.base {
Modifier.color(colorMode.toPalette().link.default)
}
SilkTheme
contains very simple (e.g. black and white) defaults, but you can override them in
an @InitSilk
method, perhaps to something that is more brand-aware:
// Assume a bunch of color constants (e.g. BRAND_LIGHT_COLOR) are defined somewhere
@InitSilk
fun overrideSilkTheme(ctx: InitSilkContext) {
ctx.theme.palettes.light.background = BRAND_LIGHT_BACKGROUND
ctx.theme.palettes.light.color = BRAND_LIGHT_COLOR
ctx.theme.palettes.dark.background = BRAND_DARK_BACKGROUND
ctx.theme.palettes.dark.color = BRAND_DARK_COLOR
}
You may find yourself occasionally wanting to define a style that should only be applied along with / after another style.
The easiest way to accomplish this is by extending the base CSS style block, using the extendedBy
method:
val GeneralTextStyle = CssStyle {
base { Modifier.fontSize(16.px).fontFamily("...") }
}
val EmphasizedTextStyle = GeneralTextStyle.extendedBy {
base { Modifier.fontWeight(FontWeight.Bold) }
}
// Or, using the `base` methods:
// val GeneralTextStyle = CssStyle.base {
// Modifier.fontSize(16.px).fontFamily("...")
// }
// val EmphasizedTextStyle = GeneralTextStyle.extendedByBase {
// Modifier.fontWeight(FontWeight.Bold)
// }
Once extended, you only need to call toModifier
on the extended style to include both styles automatically:
SpanText("WARNING", EmphasizedTextStyle.toModifier())
// You do NOT need to reference the base style, i.e.
// GeneralTextStyle.toModifier().then(EmphasizedTextStyle.toModifier())
So far, we've discussed CSS style blocks as defining a general assortment of CSS style properties. However, there is a way to define typed CSS style blocks, which let you generate typed variants associated with, and only compatible with, a specific base style.
The base style in this case is called a component style because the pattern is effective when defining widget components. In fact, it is the standard pattern that Silk uses for every single one of its widgets.
We'll discuss this full pattern around building widgets using component styles later, but to start we'll demonstrate how
to declare one. You create a marker interface that implements ComponentKind
and then pass that into your CssStyle
declaration block.
For example, if you were creating your own Button
widget:
sealed interface ButtonKind : ComponentKind
val ButtonStyle = CssStyle<ButtonKind> { /* ... */ }
Notice two points about our interface declaration:
- It is marked
sealed
. This is technically not necessary to do, but we recommend it as a way to express your intention that no one else is going to subclass it further. - The interface is empty. It is just a marker interface, useful only in enforcing typing for variants. This is discussed more in the next section.
Like normal CssStyle
declarations, the associated name is derived from its property name. You can use a @CssName
annotation to override this behavior.
The power of component styles is they can generate component variants, using the addVariant
method:
val OutlinedButtonVariant: CssStyleVariant<ButtonKind> = ButtonStyle.addVariant { /* ... */ }
Note
The recommended naming convention for variants is to take their associated style and use its name as a suffix plus the word "Variant", e.g. "ButtonStyle" -> "OutlinedButtonVariant" and "TextStyle" -> "EmphasizedTextVariant".
Important
Like a CssStyle
, your CssStyleVariant
must be public. This is for the same reason: because code gets generated
inside a main.kt
file by the Kobweb Gradle plugin, and that code needs to be able to access your variant in order to
register it.
You can technically make a variant private if you add a bit of boilerplate to handle the registration yourself:
@Suppress("PRIVATE_COMPONENT_VARIANT")
private val ExampleCustomVariant = ButtonStyle.addVariant {
/* ... */
}
// Or, `private val _ExampleCustomVariant`
@InitSilk
fun registerPrivateVariant(ctx: InitSilkContext) {
// When registering variants, using a leading dash will automatically prepend the bast style name.
// This example here will generate the final name "button-example".
ctx.theme.registerVariant("-example", ExampleCustomVariant)
}
However, you are encouraged to keep your variants public and let the Kobweb Gradle plugin handle everything for you.
The idea behind component variants is that they give the widget author power to define a base style along with one or more common tweaks that users might want to apply on top of it. (And even if a widget author doesn't provide any variants for the style, any user can always define their own in their own codebase.)
Let's revisit the button style example, bringing everything together.
sealed interface ButtonKind : ComponentKind
val ButtonStyle = CssStyle<ButtonKind> { /* ... */ }
// Note: Creates a CSS style called "button-outlined"
val OutlinedButtonVariant = ButtonStyle.addVariant { /* ... */ }
// Note: Creates a CSS style called "button-inverted"
val InvertedButtonVariant = ButtonStyle.addVariant { /* ... */ }
When used with a component style, the toModifier()
method optionally takes a variant parameter. When a variant is
passed in, both styles will be applied -- the base style followed by the variant style.
For example, ButtonStyle.toModifier(OutlinedButtonVariant)
applies the main button style first layered on top with
some additional outline styling.
You can annotate style variants with the @CssName
annotation, exactly like you can with CssStyle
. Using a leading
dash will automatically prepend the base style name. For example:
@CssName("custom-name")
val OutlinedButtonVariant = ButtonStyle.addVariant { /* ... */ } // Creates a CSS style called "custom-name"
@CssName("-custom-name")
val InvertedButtonVariant = ButtonStyle.addVariant { /* ... */ } // Creates a CSS style called "button-custom-name"
Like CssStyle.base
, variants that don't need to support additional selectors can use addVariantBase
instead to
slightly simplify their declaration:
val HighlightedCustomVariant by CustomStyle.addVariantBase {
Modifier.backgroundColor(Colors.Green)
}
// Short for
// val HighlightedCustomVariant by CustomStyle.addVariant {
// base { Modifier.backgroundColor(Colors.Green) }
// }
Silk uses component styles when defining its widgets, and you can too! The full pattern looks like this:
sealed interface CustomWidgetKind : ComponentKind
val CustomWidgetStyle = CssStyle<CustomWidgetKind> { /* ... */ }
@Composable
fun CustomWidget(
modifier: Modifier = Modifier,
variant: CssStyleVariant<CustomWidgetKind>? = null,
@Composable content: () -> Unit
) {
val finalModifier = CustomWidgetStyle.toModifier(variant).then(modifier)
/* ... */
}
In other words:
- you define a composable widget method
- it should take in a
modifier
as its first optional parameter. - it should take in an optional
CssStyleVariant
parameter (typed to your uniqueComponentKind
implementation) - inside your widget, you should apply the modifiers in order of: base style, then variant, then user modifier.
- it should end with a
@Composable
context lambda parameter (unless this widget doesn't support custom content)
A caller might call your widget one of several ways:
// Approach #1: Use default styling
CustomWidget { /* ... */ }
// Approach #2: Tweak default styling with a variant
CustomWidget(variant = TransparentWidgetVariant) { /* ... */ }
// Approach #3: Tweak default styling with inline overrides
CustomWidget(Modifier.backgroundColor(Colors.Blue)) { /* ... */ }
// Approach #4: Tweak default styling with both a variant and inline overrides
CustomWidget(Modifier.backgroundColor(Colors.Blue), variant = TransparentWidgetVariant) { /* ... */ }
In CSS, animations work by letting you define keyframes in a stylesheet which you then reference, by name, in an animation style. You can read more about them on Mozilla's documentation site.
For example, here's the CSS for an animation of a sliding rectangle (from this tutorial):
div {
width: 100px;
height: 100px;
background: red;
position: relative;
animation: mymove 5s infinite;
}
@keyframes shift-right {
from {left: 0px;}
to {left: 200px;}
}
Kobweb lets you define your keyframes in code by using a Keyframes
block:
val ShiftRightKeyframes = Keyframes {
from { Modifier.left(0.px) }
to { Modifier.left(200.px) }
}
// Later
Div(
Modifier
.size(100.px).backgroundColor(Colors.Red).position(Position.Relative)
.animation(ShiftRightKeyframes.toAnimation(
duration = 5.s,
iterationCount = AnimationIterationCount.Infinite
))
.toAttrs()
)
The name of the keyframes block is automatically derived from the property name (here, ShiftRightKeyframes
is
converted into "shift-right"
). You can then use the toAnimation
method to convert your collection of keyframes into
an animation that uses them, which you can pass into the Modifier.animation
modifier.
Important
When you declare a Keyframes
animation, it must be public. This is because code gets generated inside a main.kt
file by the Kobweb Gradle plugin, and that code needs to be able to access your variant in order to register it.
In general, it's a good idea to think of animations as global anyway, since technically they all live in a globally applied stylesheet, and you have to make sure that the animation name is unique across your whole application.
You can technically make an animation private if you add a bit of boilerplate to handle the registration yourself:
@Suppress("PRIVATE_KEYFRAMES")
private val ExampleKeyframes = Keyframes { /* ... */ }
// Or, `private val _ExampleKeyframes`
@InitSilk
fun registerPrivateAnim(ctx: InitSilkContext) {
ctx.stylesheet.registerKeyframes("example", ExampleKeyframes)
}
However, you are encouraged to keep your keyframes public and let the Kobweb Gradle plugin handle everything for you.
Occasionally, you may need access to the raw element backing the Silk widget you've just created. All Silk widgets
provide an optional ref
parameter which takes a listener that provides this information.
Box(
ref = /* ... */
) {
/* ... */
}
All ref
callbacks (discussed more below) will receive an org.w3c.dom.Element
subclass. You can check out the
Element class (and its often more
relevant HTMLElement inheritor) to see the
methods and properties that are available on it.
Raw HTML elements expose a lot of functionality not available through the higher-level Compose HTML APIs.
For a trivial but common example, we can use the raw element to capture focus:
Box(
ref = ref { element ->
// Triggered when this Box is first added into the DOM
element.focus()
}
)
The ref { ... }
method can actually take one or more optional keys of any value. If any of these keys change on a
subsequent recomposition, the callback will be rerun:
val colorMode by ColorMode.currentState
Box(
// Callback will get triggered each time the color mode changes
ref = ref(colorMode) { element -> /* ... */ }
)
If you need to know both when the element enters AND exits the DOM, you can use disposableRef
instead. With
disposableRef
, the very last line in your block must be a call to onDispose
:
val activeElements: MutableSet<HTMLElement> = /* ... */
/* ... later ... */
Box(
ref = disposableRef { element ->
activeElements.put(element)
onDispose { activeElements.remove(element) }
}
)
The disposableRef
method can also take keys that rerun the listener if any of them change. The onDispose
callback
will also be triggered in that case, as the old effect gets discarded.
And, finally, you may want to have multiple listeners that are recreated independently of one another based on different
keys. You can use refScope
as a way to combine two or more ref
and/or disposableRef
calls in any combination:
val isFeature1Enabled: Boolean = /* ... */
val isFeature2Enabled: Boolean = /* ... */
Box(
ref = refScope {
ref(isFeature1Enabled) { element -> /* ... */ }
disposableRef(isFeature2Enabled) { element -> /* ... */; onDispose { /* ... */ } }
}
)
You may occasionally want the backing element of a normal Compose HTML widget, such as a Div
or Span
. However, these
widgets don't have a ref
callback, as that's a convenience feature provided by Silk.
You still have a few options in this case.
The official way to retrieve a reference is by using a ref
block inside an attrs
block. This version of ref
is
actually more similar to Silk's disposableRef
concept than its ref
one, as it requires an onDispose
block:
Div(attrs = {
ref { element -> /* ... */; onDispose { /* ... */ } }
})
The above snippet was adapted from the official tutorials.
You could put that exact same logic inside the Modifier.toAttrs
block if you're terminating some modifier chain:
Div(attrs = Modifier.toAttrs {
ref { element -> /* ... */; onDispose { /* ... */ } }
})
Unlike Silk's version of ref
, Compose HTML's version does not accept keys. If you need this behavior and if the
Compose HTML widget accepts a content block (many of them do), you can call Silk's registerRefScope
method directly
within it:
Div {
registerRefScope(
disposableRef { element -> /* ... */; onDispose { /* ... */ } }
// or ref { element -> /* ... */ }
)
}
Kobweb supports CSS variables (also called CSS custom properties), which is a feature where you can store and retrieve
property values from variables declared within your CSS styles. It does this through a class called StyleVariable
.
Note
You can find official documentation for CSS custom properties here.
Using style variables is fairly simple. You first declare one without a value (but lock it down to a type) and later you
can initialize it within a style using Modifier.setVariable(...)
:
val dialogWidth by StyleVariable<CSSLengthNumericValue>()
// This style will be applied to a div that lives at the root, so that
// this variable value will be made available to all children.
val RootStyle = CssStyle.base {
Modifier.setVariable(dialogWidth, 600.px)
}
Tip
Compose HTML provides a CSSLengthValue
, which represents concrete values like 10.px
or 5.cssRem
. However, Kobweb
provides a CSSLengthNumericValue
type which represents the concept more generally, e.g. as the result of
intermediate calculations. There are CSS*NumericValue
types provided for all relevant units, and it is recommended
to use them when declaring style variables as they more naturally support being used in calculations.
We discuss CSSNumericValue types▼ in more detail later in this document.
You can later query variables using the value()
method to extract their current value:
val DialogStyle = CssStyle.base {
Modifier.width(dialogWidth.value())
}
You can also provide a fallback value, which, if present, would be used in the case that a variable hadn't already been set previously:
val DialogStyle = CssStyle.base {
// Will be the value of the dialogWidth variable if it was set, otherwise 500px
Modifier.width(dialogWidth.value(500.px))
}
Additionally, you can also provide a default fallback value when declaring the variable:
// Note the default fallback: 100px
val dialogWidth by StyleVariable<CSSLengthNumericValue>(100.px)
val DialogStyle100 = CssStyle.base {
// Uses default fallback. width = 100px
Modifier.width(dialogWidth.value())
}
val DialogStyle200 = CssStyle.base {
// Uses specific fallback. width = 200px
Modifier.width(dialogWidth.value(200.px))
}
val DialogStyle300 = CssStyle.base {
// Fallback (400px) ignored because variable is set explicitly. width = 300px
Modifier.setVariable(dialogWidth, 300.px).width(dialogWidth.value(400.px))
}
Caution
In the above example in the DialogStyle300
style, we set a variable and query it in the same line, which we did
purely for demonstration purposes. In practice, you would probably never do this -- the variable would have been set
separately elsewhere, e.g. in an inline style or on a parent container.
To demonstrate these concepts all together, below we declare a background color variable, create a root container scope which sets it, a child style that uses it, and, finally, a child style variant that overrides it:
// Default to a debug color, so if we see it, it indicates we forgot to set it later
val bgColor by StyleVariable<CSSColorValue>(Colors.Magenta)
val ContainerStyle = CssStyle.base {
Modifier.setVariable(bgColor, Colors.Blue)
}
val SquareStyle = CssStyle.base {
Modifier.size(100.px).backgroundColor(bgColor.value())
}
val RedSquareStyle = SquareStyle.extendedByBase {
Modifier.setVariable(bgColor, Colors.Red)
}
The following code brings the above styles together (and in some cases uses inline styles to override the background color further):
@Composable
fun ColoredSquares() {
Box(ContainerStyle.toModifier()) {
Column {
Row {
// 1: Read color from ContainerStyle
Box(SquareStyle.toModifier())
// 2: Override color via RedSquareStyle
Box(RedSquareStyle.toModifier())
}
Row {
// 3: Override color via inline styles
Box(SquareStyle.toModifier().setVariable(bgColor, Colors.Green))
Span(Modifier.setVariable(bgColor, Colors.Yellow).toAttrs()) {
// 4: Read color from parent's inline style
Box(SquareStyle.toModifier())
}
}
}
}
}
The above renders the following output:
You can also set CSS variables directly from code if you have access to the backing HTML element. Below, we use the
ref
callback to get the backing element for a fullscreen Box
and then use a Button
to set it to a random color
from the colors of the rainbow:
// We specify the initial color of the rainbow here, since the variable
// won't otherwise be set until the user clicks a button.
val bgColor by StyleVariable<CSSColorValue>(Colors.Red)
val ScreenStyle = CssStyle.base {
Modifier.fillMaxSize().backgroundColor(bgColor.value())
}
@Page
@Composable
fun RainbowBackground() {
val roygbiv = listOf(Colors.Red, /*...*/ Colors.Violet)
var screenElement: HTMLElement? by remember { mutableStateOf(null) }
Box(ScreenStyle.toModifier(), ref = ref { screenElement = it }) {
Button(onClick = {
// You can call `setVariable` on the backing HTML element to set the variable value directly
screenElement!!.setVariable(bgColor, roygbiv.random())
}) {
Text("Click me")
}
}
}
The above results in the following UI:
Most of the time, you can actually get away with not using CSS Variables! Your Kotlin code is often a more natural place to describe dynamic behavior than HTML / CSS is.
Let's revisit the "colored squares" example from above. Note it's much easier to read if we don't try to use variables at all.
val SquareStyle = CssStyle.base {
Modifier.size(100.px)
}
@Composable
fun ColoredSquares() {
Column {
Row {
Box(SquareStyle.toModifier().backgroundColor(Colors.Blue))
Box(SquareStyle.toModifier().backgroundColor(Colors.Red))
}
Row {
Box(SquareStyle.toModifier().backgroundColor(Colors.Green))
Box(SquareStyle.toModifier().backgroundColor(Colors.Yellow))
}
}
}
And the "rainbow background" example is similarly easier to read by using Kotlin variables
(i.e. var someValue by remember { mutableStateOf(...) }
) instead of CSS variables:
val ScreenStyle = CssStyle.base {
Modifier.fillMaxSize()
}
@Page
@Composable
fun RainbowBackground() {
val roygbiv = listOf(Colors.Red, /*...*/ Colors.Violet)
var currColor by remember { mutableStateOf(Colors.Red) }
Box(ScreenStyle.toModifier().backgroundColor(currColor)) {
Button(onClick = { currColor = roygbiv.random() }) {
Text("Click me")
}
}
}
Even though you should rarely need CSS variables, there may be occasions where they can be a useful tool in your toolbox. The above examples were artificial scenarios used as a way to show off CSS variables in relatively isolated environments. But here are some situations that might benefit from CSS variables:
- You have a site which allows users to choose from a list of several themes (e.g. primary and secondary colors). It
would be trivial enough to add CSS variables for
themePrimary
andthemeSecondary
(applied at the site's root) which you can then reference throughout your styles. - You need more control for colors in your theming than can be provided for by the simple light / dark color mode. For example, Wordle has light / dark + normal / contrast modes.
- You want to create a widget which dynamically changes its behavior based on the context it is added within. For example, maybe your site has a dark area and a light area, and the widget should use white outlines in the dark area and black outlines in the light. This can be accomplished by exposing an outline color variable, which each area of your site is responsible for setting.
- You want to allow the user to tweak values within a pseudo-class selector (e.g. hover, focus, active) for some widget (e.g. color or border size), which is much easier to do using variables than listening to events and setting inline styles.
- You have a widget that you ended up creating a bunch of variants for, but instead you realize you could replace them all with one or two CSS variables.
When in doubt, lean on Kotlin for handling dynamic behavior, and occasionally consider using style variables if you feel doing so would clean up the code.
Kobweb provides the silk-icons-fa
artifact which you can use in your project if you want access to all the free
Font Awesome (v6) icons.
Using it is easy! Search the Font Awesome gallery, choose an icon, and then call it using the associated Font Awesome icon composable.
For example, if I wanted to add the Kobweb-themed spider icon, I could call this in my Kobweb code:
FaSpider()
That's it!
Some icons have a choice between solid and outline versions, such as "Square" (outline and filled). In that case, the default choice will be an outline mode, but you can pass in a style enum to control this:
FaSquare(style = IconStyle.FILLED)
All Font Awesome composables accept a modifier parameter, so you can tweak it further:
FaSpider(Modifier.color(Colors.Red))
Note
When you create a project using our app
template, Font Awesome icons are included.
Kobweb provides the silk-icons-mdi
artifact which you can use in your project if you want access to all the
free Material Design icons.
Using it is easy! Search the Material Icons gallery, choose an icon, and then call it using the associated Material Design Icon composable.
For example, let's say after a search I found and wanted to use their bug report icon, I could call this in my Kobweb code by converting the name to camel case:
MdiBugReport()
That's it!
Most material design icons support multiple styles: outlined, filled, rounded, sharp, and two-tone. Check the gallery
search link above to verify what styles are supported by your icon. You can identify the one you want to use by passing
it into the method's style
parameter:
MdiLightMode(style = IconStyle.TWO_TONED)
All Material Design Icon composables accept a modifier parameter, so you can tweak it further:
MdiError(Modifier.color(Colors.Red))
Outside of pages, it is common to create reusable, composable parts. While Kobweb doesn't enforce any particular rule here, we recommend a convention that, if followed, may make it easier to allow new readers of your codebase to get around.
First, as a sibling to pages, create a folder called components. Within it, add:
- layouts - High-level composables that provide entire page layouts. Most (all?) of your
@Page
pages will start by calling a page layout function first. It's possible that you will only need a single layout for your entire site. - sections - Medium-level composables that represent compound areas inside your pages, organizing a collection of many children composables. If you have multiple layouts, it's likely sections would be shared across them. For example, nav headers and footers are great candidates for this subfolder.
- widgets - Low-level composables. Focused UI pieces that you may want to reuse all around your site. For example, a stylized visitor counter would be a good candidate for this subfolder.
If you create a markdown file under the jsMain/resources/markdown
folder, a corresponding page will be created for you
at build time, using the filename as its path.
For example, if I create the following file:
// jsMain/resources/markdown/docs/tutorial/Kobweb.md
# Kobweb Tutorial
...
this will create a page that I can then visit by going to mysite.com/docs/tutorial/kobweb
Front Matter is metadata that you can specify at the beginning of your document, like so:
---
title: Tutorial
author: bitspittle
---
...
In a following section, we'll discuss how to embed code in your markdown, but for now, know that these key / value pairs can be queried in code using the page's context:
@Composable
fun AuthorWidget() {
val ctx = rememberPageContext()
// Note: You can use `markdown!!` only if you're sure that
// this composable is called while inside a page generated
// from Markdown.
val author = ctx.markdown!!.frontMatter.getValue("author").single()
Text("Article by $author")
}
Important
If you're not seeing ctx.markdown
autocomplete, you need to make sure you depend on the
com.varabyte.kobwebx:kobwebx-markdown
artifact in your project's build.gradle
.
Within your front matter, there's a special value which, if set, will be used to render a root @Composable
that adds
the rest of your markdown code as its content. This is useful for specifying a layout for example:
---
root: .components.layout.DocsLayout
---
# Kobweb Tutorial
The above will generate code like the following:
import com.mysite.components.layout.DocsLayout
@Composable
@Page
fun KobwebPage() {
DocsLayout {
H1 {
Text("Kobweb Tutorial")
}
}
}
If you have a default root that you'd like to use in most / all of your markdown files, you can specify it in the markdown block in your build script:
// site/build.gradle.kts
kobweb {
markdown {
defaultRoot.set(".components.layout.MarkdownLayout")
}
}
Kobweb Markdown front matter supports a routeOverride
key. If present, its value will be passed into the
generated @Page
annotation (see the Route Override section▲ for valid values here).
This allows you to give your URL a name that normal Kotlin filename rules don't allow for, such as a hyphen:
# AStarDemo.md
---
routeOverride: a*-demo
---
The above will generate code like the following:
@Composable
@Page("a*-demo")
fun AStarDemoPage() { /* ... */
}
The power of Kotlin + Compose HTML is interactive components, not static text! Therefore, Kobweb Markdown support enables special syntax that can be used to insert Kotlin code.
Usually, you will define widgets that belong in their own section. Just use three triple-curly braces to insert a function that lives in its own block:
# Kobweb Tutorial
...
{{{ .components.widgets.VisitorCounter }}}
which will generate code for you like the following:
@Composable
@Page
fun KobwebPage() {
/* ... */
com.mysite.components.widgets.VisitorCounter()
}
You may have noticed that the code path in the markdown file is prefixed with a .
. When you do that, the final path
will automatically be prepended with your site's full package.
Occasionally, you may want to insert a smaller widget into the flow of a single sentence. For this case, use the
${...}
inline syntax:
Press ${.components.widgets.ColorButton} to toggle the site's current color.
Caution
Spaces are not allowed within the curly braces! If you have them there, Markdown skips over the whole thing and leaves it as text.
You may wish to add imports to the code generated from your markdown. Kobweb Markdown supports registering both global imports (imports that will be added to every generated file) and local imports (those that will only apply to a single target file).
To register a global import, you configure the markdown
block in your build script:
// site/build.gradle.kts
kobweb {
markdown {
imports.add(".components.widgets.*")
}
}
Notice that you can begin your path with a "." to tell the Kobweb Markdown plugin to prepend your site's package to it. The above would ensure that every markdown file generated would have the following import:
import com.mysite.components.widgets.*
Imports can help you simplify your Kobweb calls. Revisiting an example from just above:
# Without imports
Press ${.components.widgets.ColorButton} to toggle the site's current color.
# With imports
Press ${ColorButton} to toggle the site's current color.
Local imports are specified in your markdown's front matter (and can even affect its root declaration!):
---
root: DocsLayout
imports:
- .components.sections.DocsLayout
- .components.widgets.VisitorCounter
---
...
{{{ VisitorCounter }}}
Kobweb Markdown supports callouts, which are a way to highlight pieces of information in your document. For example, you can use them to highlight notes, tips, warnings, or important messages.
To use a callout, set the first line of some blockquoted text to [!TYPE]
, where TYPE is one of the following:
- CAUTION - Calls attention to something that the user should be extra careful about.
- IMPORTANT - Important context that the user should be aware of.
- NOTE - Neutral information that the user should notice, even when skimming.
- QUESTION - A question posed whose answer is left as an exercise to the reader.
- QUOTE - A direct quote.
- TIP - Advice that the user may find useful.
- WARNING - Information that a user should be aware of to prevent errors.
> [!NOTE]
> Lorem ipsum...
> [!QUOTE]
> Lorem ipsum...
If you'd like to change the value of the default title that shows up, you can specify it in quotes:
> [!QUESTION "Something to ponder..."]
As another example, when using quotes, you can set this to the empty string, which looks clean:
> [!QUOTE ""]
> ...
If you want to specify a label that should apply globally, you can do so by overriding the blockquote handler in your
project's build script, using the convenience method SilkCalloutBlockquoteHandler
for it:
kobweb {
markdown {
handlers.blockquote.set(SilkCalloutBlockquoteHandler(labels = mapOf("QUOTE" to "")))
}
}
Caution
Callouts are provided by Silk. If your project does not use Silk and you override the blockquote handler like this, it will generate code that will cause a compile error.
Silk provides a handful of variants for callouts.
For example, an outlined variant:
and a filled variant:
You can also combine any of the standard variants with an additional matching link variant (e.g.
LeftBorderedCalloutVariant.then(MatchingLinkCalloutVariant))
) to make it so that any hyperlinks inside the callout
will match the color of the callout itself:
If you prefer any of these styles over the default, you can set the variant
parameter in the
SilkCalloutBlockquoteHandler
, for example here we set it to the outlined variant:
kobweb {
markdown {
handlers.blockquote.set(SilkCalloutBlockquoteHandler(
variant = "com.varabyte.kobweb.silk.components.display.OutlinedCalloutVariant")
)
}
}
Of course, you can also define your own variant in your own codebase and pass that in here as well.
If you'd like to register a custom callout, this is done in two parts.
First, declare your custom callout setup in your code somewhere:
package com.mysite.components.widgets.callouts
val CustomCallout = CalloutType(
/* ... specify icon, label, and colors here ... */
)
and then register it in your build script, extending the default list of handlers (i.e. SilkCalloutTypes
) with your
custom one:
kobweb {
markdown {
handlers.blockquote.set(
SilkCalloutBlockquoteHandler(types =
SilkCalloutTypes +
mapOf("CUSTOM" to ".components.widgets.callouts.CustomCallout")
)
)
}
}
Note
As seen above, by using a leading .
, you can omit your project's group (e.g. com.mysite
). Kobweb will
automatically prepend it for you.
That's it! At this point, you can use it in your markdown:
> [!CUSTOM]
> Neat.
It can be really useful to process all markdown files during your site's build. A common example is to collect all markdown articles and generate a listing page from them.
You can actually do this using pure Gradle code, but it's common enough that Kobweb provides a convenience API, via the
markdown
block's process
callback.
You can register a callback that will be triggered at build time with a list of all markdown files in your project.
kobweb {
markdown {
process.set { markdownEntries ->
// `markdownEntries` is type `List<MarkdownEntry>`, where an entry includes the file's path, the route it will
// be served at, and any parsed front matter.
println("Processing markdown files:")
markdownEntries.forEach { entry ->
println("\t* ${entry.filePath} -> ${entry.route}")
}
}
}
}
Inside the callback, you can also call generateKotlin
and generateMarkdown
utility methods. Here is a very rough
example of creating a listing page for all blog posts in a site (found under the resources/markdown/blog
folder):
kobweb {
markdown {
process.set { markdownEntries ->
generateMarkdown("blog/index.md", buildString {
appendLine("# Blog Index")
markdownEntries.forEach { entry ->
if (entry.filePath.startsWith("blog/")) {
val title = entry.frontMatter["title"] ?: "Untitled"
appendLine("* [$title](${entry.route})")
}
}
})
}
}
}
Refer to the build script of my open source blog site and search for "process.set" to see this feature in action in a production environment.
Many developers new to web development have heard horror stories about CSS, and they might hope that Kobweb, by leveraging Kotlin and a Jetpack Compose-inspired API, means they won't have to learn it.
It's worth dispelling that illusion! CSS is inevitable.
That said, CSS's reputation is probably worse than it deserves to be. Many of its features are actually fairly straightforward and some are quite powerful. For example, you can efficiently declare that your element should be wrapped with a thin border, with round corners, casting a drop shadow beneath it to give it a feeling of depth, painted with a gradient effect for its background, and animated with an oscillating, tilting effect.
It's hoped that, once you've learned a bit of CSS through Kobweb, you'll find yourself actually enjoying it (sometimes)!
Kobweb offers enough of a layer of abstraction that you can learn CSS in a more incremental way.
First and most importantly, Kobweb gives you a Kotlin-idiomatic type-safe API to CSS properties. This is a major improvement over writing CSS in text files which fail silently at runtime.
Next, layout widgets like Box
, Column
, and Row
can get you up and running quickly with rich, complex layouts
before ever having to understand what a "flex layout" is.
Meanwhile, using CssStyle
can help you break your CSS up into smaller, more manageable
pieces that live close to the code that actually uses them, allowing your project to avoid a giant, monolithic CSS file.
(Such giant CSS files are one of the reasons CSS has an intimidating reputation).
For example, a CSS file that could easily look like this:
/* Dozens of rules... */
.important {
background-color: red;
font-weight: bold;
}
.important:hover {
background-color: pink;
}
/* Dozens of other rules... */
.post-title {
font-size: 24px;
}
/* A dozen more more rules... */
can migrate to this in Kobweb:
//------------------ CriticalInformation.kt
val ImportantStyle = CssStyle {
base {
Modifier.backgroundColor(Colors.Red).fontWeight(FontWeight.Bold)
}
hover {
Modifier.backgroundColor(Colors.Pink)
}
}
//------------------ Post.kt
val PostTitleStyle = CssStyle.base { Modifier.fontSize(24.px) }
Next, Silk provides a Deferred
composable which lets you declare code that won't get rendered until the rest of the
DOM finishes first, meaning it will appear on top of everything else. This is a clean way to avoid setting CSS z-index
values (another aspect of CSS that has a bad reputation).
And finally, Silk aims to provide widgets with default styles that look good for many sites. This means you should be able to rapidly develop common UIs without running into some of the more complex aspects of CSS.
Let's walk through an example of layering CSS effects on top of a basic element.
Tip
Two of the best learning resources for CSS properties are https://developer.mozilla.org
and https://www.w3schools.com
. Keep an eye out for these when you do a web search.
We'll create the bordered, floating, oscillating element we discussed earlier. Rereading it now, here are the concepts we need to figure out how to do:
- Create a border
- Round out the corners
- Add a drop shadow
- Add a gradient background
- Add a wobble animation
Let's say we want to create an attention grabbing "welcome" widget on our site. You can always start with an empty box, which we'll put some text in:
Box(Modifier.padding(topBottom = 5.px, leftRight = 30.px)) {
Text("WELCOME!!")
}
Create a border
Next, search the internet for "CSS border". One of the top links should be: https://developer.mozilla.org/en-US/docs/Web/CSS/border
Skim the docs and play around with the interactive examples. With an understanding of the border property now, let's use code completion to discover the Kobweb version of the API:
Box(
Modifier
.padding(topBottom = 5.px, leftRight = 30.px)
.border(1.px, LineStyle.Solid, Colors.Black)
) {
Text("WELCOME!!")
}
Round out the corners
Search for "CSS rounded corners". It turns out the CSS property in this case is called a "border radius": https://developer.mozilla.org/en-US/docs/Web/CSS/border-radius
Box(
Modifier
.padding(topBottom = 5.px, leftRight = 30.px)
.border(1.px, LineStyle.Solid, Colors.Black)
.borderRadius(5.px)
) {
Text("WELCOME!!")
}
Add a drop shadow
Search for "CSS shadow". There are a few types of CSS shadow features, but after some quick reading, we realize we want to use box shadows: https://developer.mozilla.org/en-US/docs/Web/CSS/box-shadow
After playing around with blur and spread values, we get something that looks decent:
Box(
Modifier
.padding(topBottom = 5.px, leftRight = 30.px)
.border(1.px, LineStyle.Solid, Colors.Black)
.borderRadius(5.px)
.boxShadow(blurRadius = 5.px, spreadRadius = 3.px, color = Colors.DarkGray)
) {
Text("WELCOME!!")
}
Add a gradient background
Search for "CSS gradient background". This isn't a straightforward CSS property like the previous cases, so we instead get a more general documentation page explaining the feature: https://developer.mozilla.org/en-US/docs/Web/CSS/CSS_images/Using_CSS_gradients
This case turns out to be a little trickier to ultimately find the Kotlin, type-safe equivalent, but if you dig a bit more into the CSS docs, you'll learn that a linear gradient is a type of background image.
Box(
Modifier
.padding(topBottom = 5.px, leftRight = 30.px)
.border(1.px, LineStyle.Solid, Colors.Black)
.borderRadius(5.px)
.boxShadow(blurRadius = 5.px, spreadRadius = 3.px, color = Colors.DarkGray)
.backgroundImage(linearGradient(LinearGradient.Direction.ToRight, Colors.LightBlue, Colors.LightGreen))
) {
Text("WELCOME!!")
}
Add a wobble animation
And finally, search for "CSS animations": https://developer.mozilla.org/en-US/docs/Web/CSS/CSS_animations/Using_CSS_animations
You can review the Animations▲ section above for a refresher on how Kobweb supports this feature, which
requires declaring a top-level Keyframes
block which then gets referenced inside an animation modifier:
// Top level property
val WobbleKeyframes = Keyframes {
from { Modifier.rotate((-5).deg) }
to { Modifier.rotate(5.deg) }
}
// Inside your @Page composable
Box(
Modifier
.padding(topBottom = 5.px, leftRight = 30.px)
.border(1.px, LineStyle.Solid, Colors.Black)
.borderRadius(5.px)
.boxShadow(blurRadius = 5.px, spreadRadius = 3.px, color = Colors.DarkGray)
.backgroundImage(linearGradient(LinearGradient.Direction.ToRight, Colors.LightBlue, Colors.LightGreen))
.animation(
WobbleKeyframes.toAnimation(
duration = 1.s,
iterationCount = AnimationIterationCount.Infinite,
timingFunction = AnimationTimingFunction.EaseInOut,
direction = AnimationDirection.Alternate,
)
)
) {
Text("WELCOME!!")
}
And we're done!
The above element isn't going to win any style awards, but I hope this demonstrates how much power CSS can give you in just a few declarative lines of code. And thanks to the nature of CSS, combined with Kobweb's live reloading experience, we were able to experiment with our idea incrementally.
One of our main project contributors created a site called CSS 2 Kobweb
which aims to simplify the process of converting CSS examples to equivalent Kobweb CssStyle
and/or Modifier
declarations.
Tip
CSS 2 Kobweb also supports specifying class name selectors and keyframes. For example, see what happens when you paste in the following CSS code:
.site-banner {
position: relative;
padding-left: 10px;
padding-top: 5%;
animation: slide-in 3s linear 1s infinite;
background-position: bottom 10px right;
background-image: linear-gradient(to bottom, #eeeeee, white 25px);
}
.site-banner:hover {
color: rgb(40, 40, 40);
}
@keyframes slide-in {
from {
transform: translateX(-2rem) scale(0.5);
}
to {
transform: translateX(0);
opacity: 1;
}
}
The web is full of examples of interesting CSS effects. Almost any CSS-related search will result in tons of StackOverflow answers, interactive playgrounds featuring WYSIWYG editors, and blog posts. Many of these introduce some really novel CSS examples. This is a great way to learn more about web development!
However, as the previous section demonstrated, it can sometimes be a pain to go from a CSS example to the equivalent Kobweb code. We hope that CSS 2 Kobweb can help with that.
This project is already very useful, but it's still early days. If you find cases of CSS 2 Kobweb that are incorrect, please consider filing an issue in their repository.
Hopefully this section gave you insight into how you can explore CSS APIs on your own, but if you're stuck on getting an effect working, remember you can reach out to one of the options in the connecting with us▼ section, and someone in the community can probably help!
One of Kobweb's major additions on top of Compose HTML is the export process.
This feature elevates the framework from one that produces a single-page application to one that produces a whole, navigable site. The export process takes snapshots of every page, resulting in better SEO support and a quicker initial render.
A normal development workflow will have you using kobweb run
to build up your site, and then when you're ready to
publish it, you'll kobweb export
a production version.
Let's take a moment to walk through this process in more detail, in order to demystify it.
The index.html
file of a normal Compose HTML site looks like this:
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<title>My Site Title</title>
</head>
<body>
<div id="root"></div>
<script src="mysite.js"></script>
</body>
</html>
Note
For example, you can find this exact structure recommended in the official Getting Started instructions.
What this does is declare a root element whose contents will get filled out dynamically at runtime. You can think of the
mysite.js
script at the end of the file as the seed that grows into your website.
This is very powerful, but when you build a website with this approach, you run into two major issues:
- As your codebase grows larger,
mysite.js
gets bigger and bigger, meaning a larger download is required before the site gets rendered. - Search engines have a harder time indexing your site, because they can't see the content until the JavaScript executes. Any web crawler that doesn't execute JavaScript will see a blank page.
OK, so let's add Kobweb into the mix. Here, we build a very minimal page and export our site (using kobweb export
) to
see what happens.
@Page
@Composable
fun ExampleKobwebPage() {
Text("This is a minimal example to demonstrate exporting.")
}
Exporting generates the following HTML under your kobweb/.site
folder, which I've reproduced here with a bunch of
styles elided:
<!doctype html>
<html lang="en">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
<title>My Site Title</title>
<meta content="Powered by Kobweb" name="description">
<link href="/favicon.ico" rel="icon">
<meta content="width=device-width, initial-scale=1" name="viewport">
</head>
<body>
<div id="root" style="...">
<style>...</style>
<div class="..." style="min-height: 100vh;">
This is a minimal example to demonstrate exporting.
</div>
</div>
<script src="/mysite.js"></script>
</body>
</html>
As you can see, Kobweb has filled out a bunch of extra information, although the site script it still linked to at the bottom of the file. This is important since, as mentioned earlier in this section, it contains all information necessary not just to render this page but the whole site.
In other words, you can download just this page and then continue to navigate around the site without needing to download any more files.
In short, the export process will discover all @Page
-annotated methods in your codebase and generate a snapshot of
each one. You can think of each snapshot as an SEO-friendly starting point from which you can access the rest of your
site.
In order for Kobweb exporting to be able to take a snapshot of your site, it needs to spin up a browser in headless
mode. This browser is responsible for loading the simple Compose HTML version of an index.html
page and running its
JavaScript to fill out the page. The browser will then get queried for the final html which Kobweb saves to disk.
Kobweb delegates much of this task to Microsoft's excellent Playwright framework. Hopefully this will be invisible to almost all users, but for advanced cases, it can be useful to know the technology that's running under the hood.
For custom CI/CD setups, you will at the very least need to be aware that the Kobweb export process requires a browser. For users who would like more information about this, we discuss one example in more detail later, in the GitHub Workflow for exports▼ section.
There are two flavors of Kobweb sites: static and full stack.
A static site (or, more completely, a static layout site) is one where you export a bunch of frontend files (e.g.
html
, js
, and public resources) into a single, organized folder that gets served directly by
a static website hosting provider.
In other words, you don't write a single line of server code. The server is provided for you in this case and uses a fairly straightforward algorithm - it hosts all the content you upload to it as raw, static assets.
The name static does not refer to the behavior of your site but rather that of your hosting provider solution. If someone makes a request for a page, the same response bytes get served every time (even if that page is full of custom code that allows it to behave in very interactive ways).
A full stack site is one where you write both the logic that runs on the frontend (i.e. on the user's machine) and the logic that runs on the backend (i.e. on a server somewhere). This custom server must at least serve requested files (exactly the same job that a static web hosting service does) plus it likely also defines endpoints providing custom functionality tailored to your site's needs.
For example, maybe you define an endpoint which, given a user ID and an authentication token, returns that user's profile information.
When Kobweb was first written, it only provided the full stack solution, as being able to write your own server logic enabled a maximum amount of power and flexibility. The mental model for using Kobweb during this early time was simple and clear.
However, in practice, most projects don't need the power afforded by a full stack setup. A website can give users a very clean, dynamic experience simply by writing responsive frontend logic to make it look good, e.g. with animations and delightful user interactions.
Additionally, many "Feature as a Service" solutions have popped up over the years, which can provide a ton of convenient functionality that used to require a custom server. These days, you can easily integrate auth, database, and analytics solutions all without writing a single line of backend code.
The process for exporting a bunch of files in a way that can be consumed by a static web hosting provider tends to be much faster and cheaper than using a full stack solution. Therefore, you should prefer a static site layout unless you have a specific need for a full stack approach.
Some possible reasons to use a custom server (and, therefore, a full stack approach) are:
- needing to communicate with other, private backend services in your company.
- intercepting requests as an intermediary for some third-party service where you own a very sensitive API key that you don't want to leak (such as a service that delegates to ChatGPT).
- acting as a hub to connect multiple clients together (such as a chat server).
If you aren't sure which category you fall into, then you should probably be creating a static layout site. It's much easier to migrate from a static layout site to a full stack site later than the other way around.
Both site flavors require an export.
To export your site with a static layout, use the kobweb export --layout static
command, while for full stack the command is kobweb export --layout fullstack
(or just kobweb export
since
fullstack
is the default layout as it originally was the only way).
Once exported, you can test your site by running it locally before uploading. You run a static site with
kobweb run --env prod --layout static
and a full stack site with kobweb run --env prod --layout fullstack
(or just
kobweb run --env prod
).
Sometimes, you have behavior that should run when an actual user is navigating your site, but you don't want it to run at export time. For example, maybe you offer logged-in users an authenticated experience, but you'll never have a logged-in user at export time.
You can determine if your page is being rendered as part of an export by checking the PageContext.isExporting
property.
This gives you the opportunity to manipulate the exported HTML or avoid side effects associated with page loading.
@Composable
fun AuthenticatedLayout(content: @Composable () -> Unit) {
var loggedInUser by remember { mutableStateOf<User?>(null) }
val ctx = rememberPageContext()
if (!ctx.isExporting) {
LaunchedEffect(Unit) {
loggedInUser = checkForLoggedInUser() // <- A slow, expensive method
}
}
if (loggedInUser == null) {
LoggedOutScaffold { content() }
} else {
LoggedInScaffold(user) { content() }
}
}
Dynamic routes are skipped over by the export process. After all, it's not possible to know all the possible values that could be passed into a dynamic route.
However, if you have a specific instance of a dynamic route that you'd like to export, you can configure your site's build script as follows:
kobweb {
app {
export {
// "/users/{user}/posts/{post}" has special handling for the "default" / "0" case
addExtraRoute("/users/default/posts/0", "users/index.html")
}
}
}
A static site gets exported into .kobweb/site
by default (you can configure this location in your .kobweb/conf.yaml
file if you'd like). You can then upload the contents of that folder to the static web hosting provider of your choice.
Deploying a full stack site is a bit more complex, as different providers have wildly varying setups, and some users may
even decide to run their own web server themselves. However, when you export your Kobweb site, scripts are generated for
running your server, both for *nix platforms (.kobweb/server/start.sh
) and the Windows
platform (.kobweb/server/start.bat
). If the provider you are using speaks Dockerfile, you can set ENTRYPOINT
to
either of these scripts (depending on the server's platform).
Going in more detail than this is outside the scope of this README. However, you can read my blog posts for a lot more information and some clear, concrete examples:
By default, Kobweb will automatically root every page to the KobwebApp
composable
(or, if using Silk, to a SilkApp
composable).
These perform some minimal common work (e.g. applying CSS styles) that should be present across your whole site.
This means if you register a page:
// jsMain/kotlin/com/mysite/pages/Index.kt
@Page
@Composable
fun HomePage() {
/* ... */
}
then the final result that actually runs on your site will be:
// In a generated main.kt somewhere...
KobwebApp {
HomePage()
}
It is likely you'll want to configure this further for your own application. Perhaps you have some initialization logic
that you'd like to run before any page gets run (like logic for updating saved settings into local storage). And for
many apps it's a great place to specify a full screen Silk Surface
as that makes all children beneath it transition
between light and dark colors smoothly.
In this case, you can create your own root composable and annotate it with @App
. If present, Kobweb will use that
instead of its own default. You should, of course, delegate to KobwebApp
(or SilkApp
if using Silk), as the
initialization logic from those methods should still be run.
Here's an example application composable override that I use in many of my own projects:
@App
@Composable
fun MyApp(content: @Composable () -> Unit) {
SilkApp {
val colorMode = ColorMode.current
LaunchedEffect(colorMode) { // Relaunched every time the color mode changes
localStorage.setItem("color-mode", colorMode.name)
}
// A full screen Silk surface. Sets the background based on Silk's palette and animates color changes.
Surface(SmoothColorStyle.toModifier().minHeight(100.vh)) {
content()
}
}
}
You can define at most a single @App
on your site, or else the Kobweb Application plugin will complain at build
time.
The default styles picked by browsers for many HTML elements rarely fit most site designs, and it's likely you'll want to tweak at least some of them. A very common example of this is the default web font, which if left as is will make your site look a bit archaic.
Most traditional sites overwrite styles by creating a CSS stylesheet and then linking to it in their HTML. However, if
you are using Silk in your Kobweb application, you can use an approach very similar to CssStyle
discussed above but
for general HTML elements.
To do this, create an @InitSilk
method. The context parameter includes a stylesheet
property that represents the CSS
stylesheet for your site, providing a Silk-idiomatic API for adding CSS rules to it.
Below is a simple example that sets the whole site to more aesthetically pleasing fonts than the browser defaults, one for regular text and one for code:
@InitSilk
fun initSilk(ctx: InitSilkContext) {
ctx.stylesheet.registerStyleBase("body") {
Modifier.fontFamily("Ubuntu", "Roboto", "Arial", "Helvetica", "sans-serif")
.fontSize(18.px)
.lineHeight(1.5)
}
ctx.stylesheet.registerStyleBase("code") {
Modifier.fontFamily("Ubuntu Mono", "Roboto Mono", "Lucida Console", "Courier New", "monospace")
}
}
Tip
The registerStyleBase
method is commonly used for registering styles with minimal code, but you can also use
registerStyle
, especially if you want to add some support for one or more pseudo-classes (
e.g. hover
, focus
, active
):
ctx.stylesheet.registerStyle("code") {
base {
Modifier
.fontFamily("Ubuntu Mono", "Roboto Mono", "Lucida Console", "Courier New", "monospace")
.userSelect(UserSelect.None) // No copying code allowed!
}
hover {
Modifier.cursor(Cursor.NotAllowed)
}
}
Occasionally you might find yourself with a value at build time that you want your site to know at runtime.
For example, maybe you want to specify a version based on the current UTC timestamp. Or maybe you want to read a system environment variable's value and pass that into your Kobweb site as a way to configure its behavior.
This is supported via Kobweb's AppGlobals
singleton, which is like a Map<String, String>
whose values you can set
from your project's build script using the kobweb.app.globals
property.
Let's demonstrate this with the UTC version example.
In your application's build.gradle.kts
, add the following code:
import java.time.LocalDateTime
import java.time.ZoneId
import java.time.format.DateTimeFormatter
plugins {
/* ... */
alias(libs.plugins.kobweb.application)
}
kobweb {
app {
globals.put(
"version",
LocalDateTime
.now(ZoneId.of("UTC"))
.format(DateTimeFormatter.ofPattern("yyyyMMdd.kkmm"))
)
}
}
You can then access them via the AppGlobals.get
or AppGlobals.getValue
methods:
val version = AppGlobals.getValue("version")
In your Kotlin project somewhere, it is recommended that you either add some type-safe extension methods, or you can create your own wrapper object (based on your preference):
// SiteGlobals.kt
import com.varabyte.kobweb.core.AppGlobals
// Extension method approach ---------------------
val AppGlobals.version: String
get() = getValue("version")
// Wrapper object approach -----------------------
object SiteGlobals {
val version: String = AppGlobals.getValue("version")
}
At this point, you can access this value in your site's code, say for a tiny label that would look good in a footer perhaps:
// components/widgets/SiteVersion.kt
val VersionTextStyle = CssStyle.base {
Modifier.fontSize(0.6.cssRem)
}
@Composable
fun SiteVersion(modifier: Modifier = Modifier) {
SpanText("v" + SiteGlobals.version, VersionTextStyle.toModifier().then(modifier))
}
As mentioned earlier, Silk widgets all use component styles▲ to power their look and feel.
Normally, if you want to tweak a style in select locations within your site, you just create a variant from that style:
val TweakedButtonVariant = ButtonStyle.addVariantBase { /* ... */ }
// Later...
Button(variant = TweakedButtonVariant) { /* ... */ }
But what if you want to globally change the look and feel of a widget across your entire site?
You could of course create your own composable which wraps some underlying composable with its own new style, e.g.
MyButton
which defines its own MyButtonStyle
that internally delegates to Button
. However, you'd have to be
careful to make sure all new developers who add code to your site know to use MyButton
instead of Button
directly.
Silk provides another way, allowing you to modify any of its declared styles and/or variants in place.
You can do this via an @InitSilk
method, which takes an InitSilkContext
parameter. This context provides the theme
property, which provides the following family of methods for rewriting styles and variants:
@InitSilk
fun replaceStylesAndOrVariants(ctx: InitSilkContext) {
ctx.theme.replaceStyle(SomeStyle) { /* ... */ }
ctx.theme.replaceVariant(SomeVariant) { /* ... */ }
ctx.theme.modifyStyle(SomeStyle) { /* ... */ }
ctx.theme.modifyVariant(SomeVariant) { /* ... */ }
}
Note
Technically, you can use these methods with your own site's declared styles and variants as well, but there should be no reason to do so since you can just go to the source and change those values directly. However, this can still be useful if you're using a third-party Kobweb library that provides its own styles and/or variants.
Use the replace
versions if you want to define a whole new set of CSS rules from scratch, or use the modify
versions
to layer additional changes on top of what's already there.
Caution
Using replace
on some of the more complex Silk styles can be tricky, and you may want to familiarize yourself with
the details of how those widgets are implemented before attempting to do so. Additionally, once you replace a style
in your site, you will be opting-out of any future improvements to that style that may be made in future versions of
Silk.
Here's an example of replacing ImageStyle
on a site that wants to force all images to have rounded corners and
automatically scale down to fit their container:
@InitSilk
fun replaceSilkImageStyle(ctx: InitSilkContext) {
ctx.theme.replaceStyleBase(ImageStyle) {
Modifier
.clip(Rect(cornerRadius = 8.px))
.fillMaxWidth()
.objectFit(ObjectFit.ScaleDown)
}
}
and here's an example for a site that always wants its horizontal dividers to fill max width:
@InitSilk
fun makeHorizontalDividersFillWidth(ctx: InitSilkContext) {
ctx.theme.modifyStyleBase(HorizontalDividerStyle) {
Modifier.fillMaxWidth()
}
}
Let's say you've decided on creating a full stack website using Kobweb. This section walks you through setting it up as well as introducing the various APIs for communicating to the backend from the frontend.
A Kobweb project will always at least have a JavaScript component, but if you declare a JVM target, that will be used to define custom server logic that can then be used by your Kobweb site.
It's easiest to let Kobweb do it for you. In your site's build script, make sure you've declared
configAsKobwebApplication(includeServer = true)
:
// site/build.gradle.kts
import com.varabyte.kobweb.gradle.application.util.configAsKobwebApplication
plugins {
alias(libs.plugins.kotlin.multiplatform)
alias(libs.plugins.jetbrains.compose)
alias(libs.plugins.kobweb.application)
}
/* ... */
kotlin {
configAsKobwebApplication(includeServer = true)
/* ... */
}
Important
configAsKobwebApplication(includeServer = true)
declares and sets up both js()
and jvm()
Kotlin Multiplatform targets for you. If you don't
set includeServer = true
explicitly, only the JS target will be declared.
The easy way to check if everything is set up correctly is to open your project inside IntelliJ IDEA, wait for it to
finish indexing, and check that the jvmMain
folder is detected as a module (if so, it will be given a special icon
and look the same as the jsMain
folder):
You can define and annotate methods which will generate server endpoints you can interact with. To add one:
- Define your method (optionally
suspend
able) in a file somewhere under theapi
package in yourjvmMain
source directory. - The method should take exactly one argument, an
ApiContext
. - Annotate it with
@Api
For example, here's a simple method that echoes back an argument passed into it:
// jvmMain/kotlin/com/mysite/api/Echo.kt
@Api
suspend fun echo(ctx: ApiContext) {
// ctx.req is for the incoming request, ctx.res for responding back to the client
// Params are parsed from the URL, e.g. here "/api/echo?message=..."
val msg = ctx.req.params["message"] ?: ""
ctx.res.setBodyText(msg)
}
After running your project, you can test the endpoint by visiting mysite.com/api/echo?message=hello
You can also trigger the endpoint in your frontend code by using the extension api
property added to the
kotlinx.browser.window
class:
@Page
@Composable
fun ApiDemoPage() {
val coroutineScope = rememberCoroutineScope()
Button(onClick = {
coroutineScope.launch {
println("Echoed: " + window.api.get("echo?message=hello").decodeToString())
}
}) { Text("Click me") }
}
All the HTTP methods are supported (post
, put
, etc.).
These methods will throw an exception if the request fails for any reason. Note that for every HTTP method, there's a
corresponding "try" version that will return null instead (tryPost
, tryPut
, etc.).
If you know what you're doing, you can of course always use window.fetch(...)
directly.
When you define an API route, you are expected to set a status code for the response, or otherwise it will default to
status code 404
.
In other words, the following API route stub will return a 404:
@Api
suspend fun error404(ctx: ApiContext) {
}
In contrast, this minimal API route returns an OK status code:
@Api
suspend fun noActionButOk(ctx: ApiContext) {
ctx.res.status = 200
}
Important
The ctx.res.setBodyText
method sets the status code to 200 automatically for you, which is why code in the previous
section worked without setting the status directly. Of course, if you wanted to return a different status code value
after setting the body text, you could explicitly set it right after making the setBodyText
call. For example:
ctx.res.setBodyText("...")
ctx.res.status = 201
The design for defaulting to 404 was chosen to allow you to conditionally handle API routes based on input conditions, where early aborts automatically result in the client getting an error.
A very common case is creating an API route that only handles POST requests:
@Api
suspend fun updateUser(ctx: ApiContext) {
if (ctx.req.method != HttpMethod.POST) return
// ...
ctx.res.status = 200
}
Finally, note that you can add headers to your response. A common endpoint that some servers provide is a redirect (302) with an updated URL location. This would look like:
@Api
suspend fun redirect(ctx: ApiContext) {
if (ctx.req.method != HttpMethod.GET) return
ctx.res.headers["Location"] = "..."
ctx.res.status = 302
}
Simple!
Similar to dynamic @Page
routes, you can define API routes using curly braces in the same way to
indicate a dynamic value that should be captured with some binding name.
For example, the following endpoint will capture the value "123" into a key name called
"article" when querying articles/123
:
// jvmMain/kotlin/com/mysite/api/articles/Article.kt
@Api("{}")
suspend fun fetchArticle(ctx: ApiContext) {
val articleId = ctx.req.params["article"] ?: return
// ...
}
Recall from the @Page
docs that specifying a name inside the curly braces defines the variable name used to capture
the value. When empty, as above, Kobweb uses the filename to generate it. In other words, you could explicitly specify
@Api("{article}")
for the same effect.
Once this API endpoint is defined, you just query it as you would any normal API endpoint:
coroutineScope.launch {
val articleText = window.api.get("articles/123").decodeToString()
// ...
}
Finally, astute readers might notice that (like dynamic @Page
routes) we use the same property to query dynamic route
values as well as query parameters.
Captured dynamic values will always take precedence over query parameters in the params
map. In practice, this should
never be a problem, because it would be very confusing design to write an API endpoint that got called like
articles/123?article=456
. That said, you can also use ctx.req.queryParams["article"]
to disambiguate this case if
necessary.
Kobweb also supports declaring methods that should be run when your server starts up, which is particularly useful for
initializing services that your @Api
methods can then use. These methods must be annotated with @InitApi
and must
take a single InitApiContext
parameter.
Important
If you are running a development server and change any of your backend code, causing a live reloading event, the init methods will be run again.
The InitApiContext
class exposes a mutable set property (called data
) which you can put anything into. Meanwhile,
@Api
methods expose an immutable version of data
. This allows you to initialize a service in an @InitApi
method
and then access it in your @Api
methods.
Let's demonstrate a concrete example, imagining we had an interface called Database
with a mutable
subclass MutableDatabase
that implements it and provides additional APIs for mutating the database.
The skeleton for registering and later querying such a database instance might look like this:
@InitApi
fun initDatabase(ctx: InitApiContext) {
val db = MutableDatabase()
db.createTable("users", listOf("id", "name")).apply {
addRow(listOf("1", "Alice"))
addRow(listOf("2", "Bob"))
}
db.loadResource("products.csv")
ctx.data.add<Database>(db)
}
@Api
fun getUsers(ctx: ApiContext) {
if (ctx.req.method != HttpMethod.GET) return
val db = ctx.data.get<Database>()
ctx.res.setBodyText(db.query("SELECT * FROM users").toString())
}
Kobweb servers also support persistent connections via streams. Streams are essentially named channels that maintain continuous contact between the client and the server, allowing either to send messages to the other at any time. This is especially useful if you want your server to be able to communicate updates to your client without needing to poll.
Additionally, multiple clients can connect to the same stream. In this case, the server can choose to not only send a message back to your client, but also to broadcast messages to all users (or a filtered subset of users) on the same stream. You could use this, for example, to implement a chat server with rooms.
Like API routes, API streams must be defined under the api
package in your jvmMain
source directory. By default, the
name of the stream will be derived from the file name and path that it's declared in (e.g. "api/lobby/Chat.kt" will
create a channel named "lobby/chat").
Unlike API routes, API streams are defined as properties, not methods. This is because API streams need to be a bit more flexible than routes, since streams consist of multiple distinct events: client connection, client messages, and client disconnection.
Also unlike API routes, streams do not have to be annotated. The Kobweb Application plugin can automatically detect them.
For example, here's a simple stream, declared on the backend, that echoes back any argument it receives:
// jvmMain/kotlin/com/mysite/api/Echo.kt
val echo = object : ApiStream {
override suspend fun onClientConnected(ctx: ClientConnectedContext) {
// Optional: ctx.stream.broadcast a message to all other clients that ctx.clientId connected
// Optional: Update ctx.data here, initializing data associated with ctx.clientId
}
override suspend fun onTextReceived(ctx: TextReceivedContext) {
ctx.stream.send(ctx.text)
}
override suspend fun onClientDisconnected(ctx: ClientDisconnectedContext) {
// Optional: ctx.stream.broadcast a message to all other clients that ctx.clientId disconnected
// Optional: Update ctx.data here, removing data associated with ctx.clientId
}
}
To communicate with an API stream from your site, you need to create a stream connection on the client:
@Page
@Composable
fun ApiStreamDemoPage() {
val echoStream = rememberApiStream("echo", object : ApiStreamListener {
override fun onConnected(ctx: ConnectedContext) {}
override fun onTextReceived(ctx: TextReceivedContext) {
console.log("Echoed: ${ctx.text}")
}
override fun onDisconnected(ctx: DisconnectedContext) {}
})
Button(onClick = {
echoStream.send("hello!")
}) { Text("Click me") }
}
After running your project, you can click on the button and check the console logs. If everything is working properly, you should see "Echoed: hello!" for each time you press the button.
Tip
The examples/chat
template project uses API streams to implement a very simple chat application, so you can
reference that project for a more realistic example.
The above example was intentionally verbose, to showcase the broader functionality around API streams. However, depending on your use-case, you can elide a fair bit of boilerplate.
First of all, the connect and disconnect handlers are optional, so you can omit them if you don't need them. Let's simplify the echo example:
// Backend
val echo = object : ApiStream {
override suspend fun onTextReceived(ctx: TextReceivedContext) {
ctx.stream.send(ctx.text)
}
}
// Frontend
val echoStream = rememberApiStream("echo", object : ApiStreamListener {
override fun onTextReceived(ctx: TextReceivedContext) {
console.log("Echoed: ${ctx.text}")
}
})
Additionally, if you only care about the text event, there are convenience methods for that:
// Backend
val echo = ApiStream { ctx -> ctx.stream.send(ctx.text) }
// Frontend
val echoStream = rememberApiStream("echo") {
ctx -> console.log("Echoed: ${ctx.text}")
}
In practice, your API streams will probably be a bit more involved than the echo example above, but it's nice to know that you can handle some cases only needing a one-liner on the server and another on the client to create a persistent client-server connection!
Note
If you need to create an API stream with stricter control around when it actually connects to the server, you can
create the ApiStream
object directly instead of using rememberApiStream
:
val echoStream = remember { ApiStream("echo") }
val scope = rememberCoroutineScope()
// Later, perhaps after a button is clicked...
scope.launch {
echoStream.connect(object : ApiStreamListener { /* ... */ })
}
When faced with a choice, use API routes as often as you can. They are conceptually simpler, and you can query API endpoints with a CLI program like curl and sometimes even visit the URL directly in your browser. They are great for handling queries of or updates to server resources in response to user-driven actions (like visiting a page or clicking on a button). Every operation you perform returns a clear response code in addition to some payload information.
Meanwhile, API streams are very flexible and can be a natural choice to handle high-frequency communication. But they are also more complex. Unlike a simple request / response pattern, you are instead opting in to manage a potentially long lifetime during which you can receive any number of events. You may have to concern yourself about interactions between all the clients on the stream as well. API streams are fundamentally stateful.
You often need to make a lot of decisions when using API streams. What should you do if a client or server disconnects earlier than expected? How do you want to communicate to the client that their last action succeeded or failed (and you need to be clear about exactly which action because they might have sent another one in the meantime)? What structure do you want to enforce, if any, between a client and server connection where both sides can send messages to each other at any time?
Most importantly, API streams may not horizontally scale as well as API routes. At some point, you may find yourself in a situation where a new web server is spun up to handle some intense load.
If you're using API routes, you're already probably delegating to a database service as your data backend, so this may just work seamlessly.
But for API streams, you many naturally find yourself writing a bunch of broadcasting code. However, this only works to communicate between all clients that are connected to the same server. Two clients connected to the same stream on different servers are effectively in different, disconnected worlds.
The above situation is often handled by using a pubsub service (like Redis). This feels somewhat equivalent to using a database as a service in the API route situation, but this code might not be as straightforward to migrate.
API routes and API streams are not a you-must-use-one-or-the-other situation. Your project can use both! In general, try to imagine the case where a new server might get spun up, and design your code to handle that situation gracefully. API routes are generally safe to use, so use them often. However, if you have a situation where you need to communicate events in real-time, especially situations where you want your client to be continuously directed what to do by the server via events, API streams are a great choice.
Note
You can also search online about REST vs WebSockets, as these are the technologies that API routes and API streams are implemented with. Any discussions about them should apply here as well.
It is very common to want to set a value on one page that should be made available on other pages. Or maybe you want a value to be restored when a user returns to the page again in the future, even if they've since closed and reopened the browser.
In the world of web development, this is accomplished with web storage, of which there are two flavors: local storage and session storage.
Local storage and web storage have identical APIs, with the key difference being their lifetimes. Local storage values will last until the user clears their browser's cache, while session storage will last until the user closes the current tab.
As you might expect, local storage is useful for values that should stick around indefinitely. User preferences are a common use case here. Many Kobweb sites save the user's last selected color mode in local storage, for example.
Meanwhile, session storage is useful when you want to persist data just as long as the user is interacting with your site but no longer. For example, you might keep track of values typed into text fields that haven't been submitted to the server yet, just in case the user reloads the page by accident (page reloads do not end sessions).
Using the storage APIs in Kotlin is trivial -- just reference the kotlinx.browser.localStorage
and
kotlinx.browser.sessionStorage
objects, which are both of type Storage
:
// Note: Several fields elided for simplicity...
interface Storage {
fun getItem(key: String): String?
fun setItem(key: String, value: String)
}
import kotlinx.browser.localStorage
localStorage.setItem("example-key", "example-value")
assert(localStorage.getItem("example-key") == "example-value")
With these APIs, the developer can check if an expected value is present in storage or not when visiting a page and act accordingly, for example by re-routing users to a login page if they detect the user is not logged in.
Kobweb provides the StorageKey
utility class to enable the creation and querying of type-safe storage values.
For example, if you want to store an integer value, you can do so like this:
val BRIGHTNESS_KEY = IntStorageKey("brightness")
localStorage.setItem(BRIGHTNESS_KEY, 100)
val brightness = localStorage.getItem(BRIGHTNESS_KEY) ?: 100
The StorageKey
class (and all implementors) can take an optional defaultValue
parameter, which can help reduce some
of the boilerplate in the above code:
val BRIGHTNESS_KEY = IntStorageKey("brightness", defaultValue = 100)
val brightness = localStorage.getItem(BRIGHTNESS_KEY)!!
While typed key support is provided for all primitive types (strings, booleans, ints, floats, etc.) and also enums, you can create your own custom implementations by providing your own to-string and from-string conversion functions:
class User(val name: String, val id: String)
class UserStorageKey(name: String) : StorageKey<User>(name) {
override fun convertToString(value: User) = "$name:$id"
override fun convertFromString(value: String): User? = value.split(":")
.takeIf { it.size == 2}
?.let { User(it[0], it[1]) }
}
val LOGGED_IN_USER_KEY = UserStorageKey("logged-in-user")
val loggedInUser = localStorage.getItem(LOGGED_IN_USER_KEY)
If you are using Kotlinx serialization in your project, you can use it to simplify the above code:
@Serializable
class User(val name: String, val id: String)
class UserStorageKey(name: String) : StorageKey<User>(name) {
override fun convertToString(value: User): String = Json.encodeToString(value)
override fun convertFromString(value: String): User? =
try { Json.decodeFromString(value) } catch (ex: Exception) { null }
}
For simplicity, new projects can choose to put all their pages and widgets inside a single application module, e.g.
site/
.
However, you can define components and/or pages in separate modules and apply the com.varabyte.kobweb.library
plugin
on them (in contrast to your main module which applies the com.varabyte.kobweb.application
plugin.)
In other words, you can split up and organize your project like this:
my-project
├── sitelib
│ ├── build.gradle.kts # apply "com.varabyte.kobweb.library"
│ └── src/jsMain
│ └── kotlin.org.example.myproject.sitelib
│ ├── components
│ └── pages
└── site
├── build.gradle.kts # apply "com.varabyte.kobweb.application"
├── .kobweb/conf.yaml
└── src/jsMain
└── kotlin.org.example.myproject.site
├── components
└── pages
If you'd like to explore a multimodule project example, you can do so by running:
$ kobweb create examples/chat
which demonstrates a chat application with its auth and chat functionality each managed in their own separate modules.
Web workers are a standard web technology that allow you to run JavaScript code in a separate thread from your main application. Although JavaScript is famously single-threaded, web workers offer a way for you to run potentially expensive code in parallel to your main site without slowing it down.
A web worker script is entirely isolated from your main site and has no access to the DOM. The only way to communicate between them is via message passing.
Note
Astute readers may recognize the actor model here, which is an effective way to allow concurrency without worrying about common synchronization issues that plague common lock-based approaches.
A somewhat forced but easy-to-understand example of a web worker is one that computes the first N prime numbers.
While the worker is crunching away on intensive calculations, your site still works as normal, fully responsive. When the worker is finished, it posts a message to the application, which handles it by updating relevant UI elements.
Kobweb aims to make using web workers as easy as possible.
Here's everything you have to do (we'll show examples of these steps shortly):
- Create a new module and apply the Kobweb Worker Gradle plugin on it.
- Tag the
kotlin { ... }
block in your build script with aconfigAsKobwebWorker()
call.- (Optional but recommended) Specify a name for your worker. Otherwise, the generic name "worker" (with a short random suffix) will be used, which is functional but may make it harder to debug if something goes wrong.
- Declare a dependency on
"com.varabyte.kobweb:kobweb-worker"
. - Implement the
WorkerFactory
interface, providing aWorkerStrategy
that represents the core logic of your worker.
import com.varabyte.kobweb.gradle.worker.util.configAsKobwebWorker
plugins {
alias(libs.plugins.kotlin.multiplatform)
alias(libs.plugins.kobweb.worker) // or id("com.varabyte.kobweb.worker")
}
group = "example.worker"
version = "1.0-SNAPSHOT"
kotlin {
configAsKobwebWorker("example-worker")
sourceSets {
jsMain.dependencies {
implementation(libs.kobweb.worker) // or "com.varabyte.kobweb:kobweb-worker"
}
}
}
The WorkerFactory
interface is minimal:
interface WorkerFactory<I, O> {
fun createStrategy(postOutput: OutputDispatcher<O>): WorkerStrategy<I>
fun createIOSerializer(): IOSerializer<I, O>
}
This concise interface still captures a lot of information. It declares:
- What types your worker accepts as input and output messages.
- How it serializes those input and output messages.
- The strategy for handling input messages from the application.
- A
postOutput
object that can be used to send output messages back to the application.
The WorkerStrategy
class represents the core logic your worker does after receiving input from the application, as
well as exposes a self
parameter
that provides useful worker functionality.
abstract class WorkerStrategy<I> {
protected val self: DedicatedWorkerGlobalScope
abstract fun onInput(inputMessage: InputMessage<I>)
}
The OutputDispatcher
is a simple class which allows you to send output messages back to the application.
class OutputDispatcher<O> {
operator fun invoke(output: O, transferables: Transferables = Transferables.Empty)
}
// `postOutput: OutputDispatcher<String>` can be called like a normal method, e.g. `postOutput("hello!")`
Note
Do not worry about the Transferables
parameter for now. Transferable objects are a somewhat niche,
performance-related feature, and they will be discussed later. It is not expected that a majority of workers will
require them.
Finally, IOSerializer
is responsible for marshalling objects between the worker and the application.
interface IOSerializer<I, O> {
fun serializeInput(input: I): String
fun deserializeInput(input: String): I
fun serializeOutput(output: O): String
fun deserializeOutput(output: String): O
}
This class allows you to use the serialization library of your choice. However, as you'll see later, this can be a one-liner for developers using Kotlinx Serialization.
Once the Kobweb Worker Gradle plugin finds your worker factory implementation, it will generate a simple Worker
class
that wraps it.
// Generated code!
class Worker(val onOutput: WorkerContext.(O) -> Unit) {
fun postInput(input: I, transferables: Transferables = Transferables.Empty)
fun terminate()
}
Applications will interact with this Worker
and not the WorkerStrategy
directly. In fact, you should make your
worker factory implementation internal
to prevent applications from seeing anything but the worker.
You should think of the WorkerStrategy
as representing implementation details while the Worker
class represents a
public API. In other words, the WorkerStrategy
receives inputs, processes data, and posts outputs, while the Worker
allows users to post inputs and get notified when outputs are ready.
An application module (i.e. one that applies the Kobweb Application Gradle plugin) will automatically discover any
Kobweb worker dependencies, extracting its worker script and putting it under the public/
folder of your final site.
The following sections introduce concrete worker factories, which should help solidify the abstract concepts introduced above.
The simplest worker strategy possible is one that blindly repeats back whatever input it receives.
This is never a worker strategy that you'd actually create -- there wouldn't be a need for it -- but it's a good starting point for seeing a worker factory in action.
When you have a worker strategy that works with raw strings like this one does, you can use a one-line helper method to
implement the createIOSerializer
method, called createPassThroughSerializer
(since it just passes the raw strings
unmodified).
// Worker module
internal class EchoWorkerFactory : WorkerFactory<String, String> {
override fun createStrategy(postOutput: OutputDispatcher<String>) = WorkerStrategy<String> { input ->
postOutput(input)
}
override fun createIOSerializer() = createPassThroughSerializer()
}
Based on that implementation, a worker called EchoWorker
will be auto-generated at compile time. Using it in your
application looks like this:
// Application module
val worker = rememberWorker {
EchoWorker { message -> println("Echoed: $message") }
}
// Later
worker.postInput("hello!") // After a round trip: "Echoed: hello!"
That's it!
Important
Note the use of the rememberWorker
method. This internally calls a remember
but also sets up disposal logic that
terminates the worker when the composable is exited. If you just use a normal remember
block, the worker may keep
running longer than you expect, even if you navigate to another part of your site.
You can also stop a worker yourself by calling worker.terminate()
directly.
This next worker strategy will take in an Int
value from the user. This number represents how many seconds to count
down, firing a message for each second that passes.
This is another strategy that you'd never need in practice -- you'd just use the window.setInterval
method yourself
in your site script -- but we'll show this anyway to demonstrate two additional concepts on top of the echo worker:
- How to define a custom message serializer.
- The fact that you can call
postOutput
as often as you want.
// Worker module
internal class CountDownWorkerFactory : WorkerFactory<Int, Int> {
override fun createStrategy(postOutput: OutputDispatcher<Int>) = WorkerStrategy<Int> { input ->
var nextCount = input
var intervalId: Int = 0
intervalId = self.setInterval({ // A
postOutput(nextCount) // B
if (nextCount > 0) { --nextCount } else { self.clearInterval(intervalId) }
}, 1000)
}
override fun createIOSerializer() = object : IOSerializer<Int, Int> { // C
override fun serializeInput(input: Int) = input.toString()
override fun deserializeInput(input: String) = input.toInt()
override fun serializeOutput(output: Int) = output.toString()
override fun deserializeOutput(output: String) = output.toInt()
}
}
Notice the three comment tags above.
- A: We use
self.setInterval
(andself.clearInterval
later) instead of thewindow
object to do this. This is because thewindow
object is only available in the main script and using it here will throw an exception. - B: You can use
postOutput
any time following an input message, not just in direct response to one. - C: This is how you define a custom message serializer. You shouldn't worry about receiving improperly formatted
strings in your
deserialize
calls, because you control them! In other words, the only way you'd get a bad string is if you generated it yourself in either of theserialize
methods. If a message serializer ever does throw an exception, then the Kobweb worker will simply ignore it as a bad message.
Using the worker in your application looks like this:
// Application module
val worker = rememberWorker {
CountDownWorker {
if (it > 0) {
console.log(it + "...")
} else {
console.log("HAPPY NEW YEAR!!!")
}
}
}
// Later
worker.postInput(10) // 10... 9... 8... etc.
Tip
If you need really accurate, consistent interval timers, creating a worker like this may actually be beneficial. According to this article, web worker timers are slightly more accurate than timers run in the main thread, as they don't have to compete with the rest of the site's responsibilities. Also, it seems that web workers timers stay consistent even if the site tab loses focus.
Finally, we get to the worker idea we introduced in the very first section -- finding the first N primes.
This kind of worker likely looks like one that would actually get used in a real codebase, that being a worker which performs a potentially expensive, UI-agnostic calculation.
We'll also use this example to demonstrate how to use Kotlinx Serialization to easily declare rich input and output message types.
First, add kotlinx-serialization
and kobwebx-serialization-kotlinx
to your dependencies:
// build.gradle.kts
kotlin {
configAsKobwebWorker()
jsMain.dependencies {
implementation(libs.kotlinx.serialization.json) // or "org.jetbrains.kotlinx:kotlinx-serialization-json"
implementation(libs.kobwebx.worker.kotlinx.serialization) // or "com.varabyte.kobwebx:kobwebx-serialization-kotlinx"
}
}
Then, define the worker factory:
@Serializable
data class FindPrimesInput(val max: Int)
@Serializable
data class FindPrimesOutput(val max: Int, val primes: List<Int>)
private fun findPrimes(max: Int): List<Int> {
// Loop through all numbers, taking out multiples of each prime
// e.g. 2 will take out 4, 6, 8, 10, etc.
// then 3 will take out 9, 15, 21, etc. (6, 12, and 18 were already removed)
val primes = (1..max).toMutableList()
var primeIndex = 1 // Skip index 0, which is 1.
while (primeIndex < primes.lastIndex) {
val prime = primes[primeIndex]
var maybePrimeIndex = primeIndex + 1
while (maybePrimeIndex <= primes.lastIndex) {
if (primes[maybePrimeIndex] % prime == 0) {
primes.removeAt(maybePrimeIndex)
} else {
++maybePrimeIndex
}
}
primeIndex++
}
return primes
}
internal class FindPrimesWorkerFactory: WorkerFactory<FindPrimesInput, FindPrimesOutput> {
override fun createStrategy(postOutput: OutputDispatcher<FindPrimesOutput>) =
object : WorkerStrategy<FindPrimesInput>() {
override fun onInput(inputMessage: InputMessage<FindPrimesInput>) {
val input = inputMessage.input
postOutput(FindPrimesOutput(input.max, findPrimes(input.max)))
}
}
override fun createIOSerializer() = Json.createIOSerializer<FindPrimesInput, FindPrimesOutput>()
}
Most of the complexity above is the findPrimes
algorithm itself!
The onInput
handler is about as easy as it gets. Notice that we pass the input max
value back into the output, so
that the receiving application can easily correlate the output with the input.
And finally, note the use of the Json.createIOSerializer
method call. This utility method comes from the
kobwebx-serialization-kotlinx
dependency, allowing you to use a one-liner to implement all the serialization methods
for you.
Tip
It's fairly trivial to write the message serializer yourself if you don't want to pull in the extra dependency (or if you are using a different serialization library):
object : IOSerializer<FindPrimesInput, FindPrimesOutput> {
override fun serializeInput(input: FindPrimesInput): String = Json.encodeToString(input)
override fun deserializeInput(input: String): FindPrimesInput = Json.decodeFromString(input)
override fun serializeOutput(output: FindPrimesOutput): String = Json.encodeToString(output)
override fun deserializeOutput(output: String): FindPrimesOutput = Json.decodeFromString(output)
}
Using the worker in your application looks like this:
// Application module
val worker = rememberWorker {
FindPrimesWorker {
println("Primes for ${it.max}: ${it.primes}")
}
}
// Later
worker.postInput(FindPrimesInput(1000)) // Primes for 1000: [1, 2, 3, 5, 7, 11, ..., 977, 983, 991, 997]
The richly-typed input and output messages allow for a very explicit API here, and in the future, more parameters could be added (with default values) to either input or output classes, extending the functionality of your workers without breaking existing code.
We don't show it here, but you could also create sealed classes for your input and output messages, allowing you to define multiple types of messages that your worker can receive and respond to.
Occasionally, you may find yourself with a very large blob of data in your main application that you want to pass to a worker (or vice versa!). For example, maybe your worker will be responsible for processing a potentially large, multi-megabyte image.
Serializing a large amount of data can be expensive! In fact, you may find that even though your worker can run efficiently on a background thread, sending a large amount of data to it can cause your site to experience a significant pause during the copy. This can easily be seconds if the data is large enough!
This isn't just an issue with Kobweb. This was originally a problem with standard web APIs. To support this use-case, web workers introduced the concept of transferable objects.
Instead of an object being copied over, its ownership is transferred over from one thread to another. Attempts to use the object in the original thread after that point will throw an exception.
Kobweb workers support transferable objects in a type-safe, Kotlin-idiomatic way, via the Transferables
class. Using
it, you can register named objects in one thread and then retrieve them by that name in another.
Here's an example where we send a very large array over to a worker.
// In your site:
val largeArray = Uint8Array(1024 * 1024 * 8).apply { /* initialize it */ }
worker.postInput(WorkerInput(), Transferables {
add("largeArray", largeArray)
})
// In the worker:
val largeArray = transferables.getUint8Array("largeArray")!!
And, of course, workers can send transferable objects back to the main application as well.
// In the worker:
val largeArray = Uint8Array(1024 * 1024 * 8).apply { /* initialize it */ }
postOutput(WorkerOutput(), Transferables {
add("largeArray", largeArray)
})
// In your site:
val worker = rememberWorker {
ExampleWorker {
val largeArray = transferables.getUint8Array("largeArray")!!
// ...
}
}
Finally, it's worth noting that not every object can be transferred. In fact, very few can! You can
refer to the official docs for
a full list of supported transferable objects.
When building a Transferables
object, the add
method is type-safe, meaning you cannot add an object that cannot then
be transferred over.
Caution
Kotlin/JS does not support a majority of the classes listed here, so neither does Kobweb as a result. If you find yourself needing one of these missing classes, consider filing an issue and we might wrap the JavaScript class into Kobweb directly and update the Transferables API.
Despite official limitations, Kobweb actually offers support for a few additional types, as a convenience. If it is possible to extract transferable content from an object, transfer that, and then build the original object back up on the other end, we are happy to do that for you.
Typed arrays, such as Int8Array
, are a great example. They are actually not transferable! Only their internal
ArrayBuffer
is.
However, when you ask Kobweb to transfer a typed array, it will instead transfer its contents for you and regenerate the outer array seamlessly on the other end. This is just boilerplate code that you would have had to write yourself anyway.
Tip
The examples/imageprocessor
template demonstrates workers leveraging Transferables
to pass image data from the
main thread to a worker and back, so you can reference that project for a complete, working example.
Due to the fundamental design of web workers, you can only define a single worker per module. If you need multiple workers, you must create multiple modules, each providing their own separate worker strategy.
The Kobweb Worker Gradle plugin will complain if it finds more than one worker factory implemented in a module.
By default, the Kobweb Worker Gradle plugin requires your worker factory class to be suffixed with WorkerFactory
so it
has guidance on how to name the final worker (for example, MyExampleWorkerFactory
would generate a worker called
MyExampleWorker
, placing it in the same package as the factory class).
// ❌ The Kobweb Worker Gradle plugin will complain about this name!
internal class MyWorkerInfoProvider : WorkerFactory<I, O> { /* ... */ }
If you don't like this constraint, you can override the kobweb.worker.fqcn
property in your build script to provide
a worker name explicitly:
// build.gradle.kts
kobweb {
worker {
fqcn.set("com.mysite.MyWorker")
}
}
at which point, you are free to name your worker factory whatever you like.
If you want to just change the name of your worker, you can omit the package:
// build.gradle.kts
kobweb {
worker {
fqcn.set(".MyWorker") // Uses the same package as the worker factory
}
}
In practice, almost every site can get away without ever using a worker, especially in Kotlin/JS where you can leverage coroutines as a way to mimic concurrency in your single-threaded site.
That said, if you know your site is going to run some logic that is not concerned with the web DOM at all, and which might additionally take a long time to run, separating that out into its own worker can be a sensible approach.
By isolating your logic into a separate worker, you not only keep it from potentially freezing your UI, but you also guarantee that it will be strongly decoupled from the rest of your site, preventing future developers from introducing potential spaghetti code issues in the future.
Another interesting use-case for a worker is isolating some sort of complex state management, where encapsulating that complexity keeps the rest of your site easier to reason about.
For example, maybe you're making a web game, and you decide to create a worker to manage all the game logic. You could of course create a Kobweb library for the same effect, but using a worker has a stronger guarantee that the logic will never interact directly with your site's UI.
Caution
You should be aware that, since a web worker is a whole separate standalone script, it needs to include its own copy of the Kotlin/JS runtime, even though your main site already has its own copy.
Even after running a dead-code elimination pass, I found that the trivial echo worker's final output was about 200K (which compressed down to 60K before being sent over the wire).
For most practical use-cases, a 60K download is not a deal-breaker, especially as most images are many multiples larger than that. But developers should be aware of this, and if this is indeed a concern, you may need to avoid using Kobweb workers on your site.
Typically, sites live at the top level. This means if you have a root file index.html
and your site is hosted at
the domain https://mysite.com
then that HTML file can be accessed by visiting https://mysite.com/index.html
.
However, in some cases your site may be hosted under a subfolder, such as https://example.com/products/myproduct/
, in which case your site's root index.html
file would live at https://example.com/products/myproduct/index.html
.
Kobweb needs to know about this subfolder structure so that it takes them into account in its routing logic. This can be
specified in your project's .kobweb/conf.yaml
file with the routePrefix
value under the site
section:
site:
title: "..."
routePrefix: "..."
where the value of routePrefix
is the part between the origin part of the URL and your site's root. For example, if
your site is rooted at https://example.com/products/myproduct/
, then the value of routePrefix
would be products/myproduct
.
Note
If you are planning to host your site on GitHub Pages using the default github.io
domain, you will need to set an
appropriate routePrefix
value. For a concrete example of setting routePrefix
for GitHub Pages,
check out this relevant section from my blog post about
exporting static layout sites.
Outside of setting your routePrefix
in the conf.yaml
file, you can design your site without explicitly mentioning
it, as Kobweb's composables handle it for you. For example, Link("docs/manuals/v123.pdf")
(or Anchor
if you're not
using Silk) will automatically resolve to https://example.com/products/myproduct/docs/manuals/v123.pdf
.
If you do need to access the route prefix in your own code, you can do so by referencing the RoutePrefix.value
companion property or by utilizing the RoutePrefix.prepend(...)
companion method. This should rarely be required in
practice, however.
Note
This note is included for anyone who wants to better understand the reason for this feature's design in Kobweb.
Normally, websites are flexible enough to be hosted under any subfolder because they can use relative paths.
For example, if you are on page a/b/c
and you need to get to e/f/g
, you would set your link to ../../../e/f/g
and not /e/f/g
. In this setup, if you referenced an icon in your top-level index.html
file, you would use
favicon.ico
(a relative path to a sibling file) and not /favicon.ico
.
However, Kobweb is built on top of a reactive framework (Compose HTML) normally meant for SPAs (Single Page Applications). In this setup, the same code is used to handle any URL in your site, at which point it intercepts the URL value and renders the associated content using some giant switch case in its routing logic.
This means that Kobweb has a single, global index.html
which is used for all pages on the site. (If you look at it,
you'll mainly see a minimal DOM skeleton that acts as a container for your dynamically generated pages, plus a script
link to the code of your site.)
Now, imagine if this index.html
linked to favicon.ico
, as in the relative path case. If you were to
visit https://example.com/products/myproduct/a/b/c
, you would ultimately get served the global index.html
file that would then look for the icon file at https://example.com/products/myproduct/a/b/c/favicon.ico
, where
it doesn't exist. In other words, Kobweb needs help to know that the icon file it is looking for is always at
https://example.com/products/myproduct/favicon.ico
, so that the index.html
file link is valid in whatever
context it is served.
If you were creating your site the traditional way, you might have one hand-crafted HTML file per page, where each one
would have its own unique link to favicon.ico
. Perhaps the rooted index.html
file would use favicon.ico
and the
one located under a/b/c
would use ../../../favicon.ico
.
But, in Kobweb, since we're using the same index.html
file for every page, we use the absolute path /favicon.ico
,
or, in the case of a route prefix being set, /${routePrefix}/favicon.ico
.
Over the lifetime of a site, you may find yourself needing to change its structure. Perhaps you need to move a handful of pages under a new folder, or you need to rename a page, etc.
However, if your site has been live for a while, you may have a ton of internal links to those pages. Worse, the rest of the web (say, Google search results, or blogs and articles) may be full of links to those old locations, so even if you can find and fix up everything on your end, you can't control what others have done.
The web has long supported the concept of redirects to handle this. By advertising what links you've changed publicly, search indices can be updated and even if someone visits your page at the old location, your server can automatically tell your browser where they should have gone instead.
In Kobweb, you can define redirects in your project's .kobweb/conf.yaml
file. You simplify define a series of from
and to
values in the server.redirects
block.
server:
redirects:
- from: "/old-page"
to: "/new-page"
Kobweb servers will pick up these redirect values from the conf.yaml
file and will intercept any matching incoming
route requests, sending back a 301 status code to the
client.
So, in the above example, if a user tries to visit https://example.com/old-page
, they will be redirected to
https://example.com/new-page
automatically. Any internal links on your site that reference the old page will also be
handled -- trying to navigate to the old location will automatically end up at the new one.
The Kobweb redirect feature also supports using regexes in the from
value, which can then be referenced in the to
section using $1
, $2
, etc. variables which will be substituted with text matches in parentheses.
Group matching can be really useful if you want to redirect a whole section of your site to a new location. For example, the following redirect rule can help if you've moved all pages from an old parent folder into a new one:
server:
redirects:
- from: "/socials/facebook/([^/]+)"
to: "/socials/meta/$1"
The last thing to note is that if you have multiple redirects, they will be processed in order and all applied. This should rarely matter in most cases, but you can use it if you need to combine both changing a folder name AND a page name:
server:
redirects:
- from: "/socials/facebook/([^/]+)"
to: "/socials/meta/$1"
- from: "(/socials/meta)/about-facebook"
to: "$1/about-meta"
Important
If you are using a third-party static hosting provider to host your site, they will be unaware of the Kobweb
conf.yaml
file, so you will need to read their documentation to learn how to configure your redirects with them.
In this case, you may be able to skip defining redirects in your own Kobweb configuration file, since it may be redundant at that point. However, it may still be useful to do for documentation purposes and to ensure you won't 404 due to an old, internal link that you forgot to update.
CSS Layers are a very powerful but also relatively new CSS feature, which allow wrapping CSS style rules inside named layers as a way to control their priorities. In short, CSS layers are arbitrary names that you specify the order of. This can be an especially useful tool when dealing with CSS style rules that are fighting with each other.
Compose HTML does not support CSS layers, but Silk does! Even if you never use layers directly in your own project, Silk uses them, so users can still benefit from the feature.
By default, Silk defines six layers (from lowest to highest ordering priority):
- reset
- base
- component-styles
- component-variants
- restricted-styles
- general-styles
The reset layer is useful for defining CSS rules that exist to compensate for browser defaults that are inconsistent with each other or to override values that exist for legacy reasons that modern web design has moved away from.
The base layer is actually not used by Silk (this may change someday), but it is provided so users can home general
CSS styles in there if they want to. It is a useful place to define global styles that should get easily overridden by
any other CSS rule defined elsewhere in your project, such as inside a CssStyle
.
The next four styles are associated with the various flavors of CssStyle
definitions:
interface SomeKind : ComponentKind
val SomeStyle = CssStyle<SomeKind> { /* ... */ } // "component-styles"
val SomeVariant = SomeStyle.addVariant { /* ... */ } // "component-variants"
class ButtonSize(/*...*/) : CssStyle.Base(/*...*/) // "restricted-styles"
val GeneralStyle = CssSTyle { /* ... */ } // "general-styles"
We chose this order to ensure that CSS styles are layered in ways that match intuition; for example, a style's variant will always layer on top of the base style itself; meanwhile, a user's declared style will always layer over a component style defined by Silk.
You can register your own custom layers inside an @InitSilk
method, using the cssLayers
property:
@InitSilk
fun initSilk(ctx: InitSilkContext) {
ctx.stylesheet.cssLayers.add("theme", "layout", "utilities")
}
When declaring new layers, you can anchor them relative to existing layers. This is useful, for example, if you want to
insert layers between Silk's base layer and its CssStyle
layers:
@InitSilk
fun initSilk(ctx: InitSilkContext) {
ctx.stylesheet.cssLayers.add("third-party", after = SilkLayer.BASE)
}
If you need to affect the layer for a CssStyle
block, you can tag it with the @CssLayer
annotation:
@CssLayer("important")
val ImportantStyle = CssStyle { /* ... */ }
Note
You should always explicitly register your layers. So, for the code above, you should also declare:
@InitSilk
fun initSilk(ctx: InitSilkContext) {
ctx.stylesheet.cssLayers.add("important")
}
If you don't do this, the browser will append add any unknown layer to the end of the CSS layer list (i.e. the highest priority spot). In many cases this will be fine, but being explicit both expresses your intention clearly and reduces the chance of your site breaking in subtle ways when a future developer adds a new layer in the future.
Silk will print out a warning to the console if it detects any unregistered layers.
@InitSilk
blocks let you register general CSS styles. You can wrap them insides layers using layer
blocks:
@InitSilk
fun initSilk(ctx: InitSilkContext) {
ctx.stylesheet.apply {
cssLayers.add("headers")
layer("headers") {
registerStyle("h1") { /* ... */ }
registerStyle("h2") { /* ... */ }
}
}
}
Of course, you can associate styles with existing layers, such as the base layer we mentioned a few sections above:
@InitSilk
fun initSilk(ctx: InitSilkContext) {
ctx.stylesheet.apply {
layer(SilkLayer.BASE) {
registerStyle("div") { /* ... */ }
registerStyle("span") { /* ... */ }
}
}
}
Finally, if you are working with third party CSS stylesheets, it can be a very useful trick to wrap them in their own layer.
For example, let's say you are fighting with a third party library whose styles are a bit too aggressive and are interfering with your own styles.
First, inside your build script, import the stylesheet using
an @import
directive:
// BEFORE
kobweb.app.index.head.add {
link {
rel = "stylesheet"
href = "/highlight.js/styles/dracula.css"
}
}
// AFTER
kobweb.app.index.head.add {
style {
unsafe {
raw("@import url(\"/highlight.js/styles/dracula.css\") layer(highlightjs);")
}
}
}
Then, register your new layer in an @InitSilk
block.
@InitSilk
fun initSilk(ctx: InitSilkContext) {
// Layer(s) referenced in build.gradle.kts
ctx.stylesheet.cssLayers.add("highlightjs", after = SilkLayer.BASE)
}
You've just tamed some wild CSS styles, congratulations!
Kobweb used to support a feature called legacy routes (you can read more about the feature here using an earlier version of the Kobweb README). This was an emergency feature added to give users time to respond to us fixing a long-standing mistake with our initial route naming algorithm without breaking their existing sites.
As of v0.18.3, we believe enough time has passed, and legacy route support has finally been removed. As a result, users who never migrated away from the feature might be seeing an error pointing them to this section.
If that is you, please follow these steps:
- In your site directory, run
../gradlew kobwebListRoutes
and look for any routes that have hyphens in them. - Search through your codebase to see if there are any versions of those links but with hyphens removed. If so, update them.
- Consider updating the
redirects
section of theconf.yaml
file to explicitly redirect from the old route to the new one. See the Redirects▲ section for more information. - If you are using a third-party hosting provider for serving your site, check their documentation to learn how to notify them of these redirects.
- Delete the
legacyRouteRedirectStrategy = ...
line from thekobweb.app
block in your build script.
Tip
This target commit demonstrates how I upgraded my blog site (which uses Firebase) to move away from Kobweb legacy route redirecting.
Occasionally, you might find yourself wanting code for your site that is better generated programmatically than written by hand.
The recommended best practice is to create a Gradle task that is associated with its own unique output directory, use the task to write some code to disk under that directory, and then add that task as a source directory for your project.
Note
The reason to encourage tasks with their own unique output directory is because this approach is very friendly with Gradle caching. You may read more here to learn about this in more detail.
Adding your task as a source directory ensures it will get triggered automatically before the Kobweb tasks responsible for processing your project are themselves run.
You want to do this even if you only plan to generate a single file. This is because associating your task with an output directory is what enables it to be used in place of a source directory.
The structure for this approach generally looks like this:
// e.g. site/build.gradle.kts
val generateCodeTask = tasks.register("generateCode") {
group = "myproject"
// You may not need an input file or dir for your task, and if so, you can exclude the next line. If you do need one,
// I'm assuming it is a data file or files in your resources somewhere.
val resInputDir = layout.projectDirectory.dir("src/jsMain/resources")
// $name here to create a unique output directory just for this task
val genOutputDir = layout.buildDirectory.dir("generated/$group/$name/src/jsMain/kotlin")
inputs.dir(resInputDir).withPathSensitivity(PathSensitivity.RELATIVE)
outputs.dir(genOutputDir)
doLast {
genOutputDir.get().file("org/example/pages/SomeCode.kt").asFile.apply {
parentFile.mkdirs()
// find and parse file out of resInputDir and write generated code here:
writeText(/* ... */)
println("Generated $absolutePath")
}
}
}
kotlin {
configAsKobwebApplication()
commonMain.dependencies { /* ... */ }
jsMain {
kotlin.srcDir(generateCodeTask) // <----- Set your task here
dependencies { /* ... */ }
}
}
In case you want to generate resources that end up in your final site as files (e.g. mysite.com/rss.xml
) and not
code, the main change you need to make is migrating the line kotlin.srcDir
to resources.srcDir
:
// e.g. site/build.gradle.kts
val generateResourceTask = tasks.register("generateResource") {
group = "myproject"
// $name here to create a unique output directory just for this task
val genOutputDir = layout.buildDirectory.dir("generated/$group/$name/src/jsMain/resources")
outputs.dir(genOutputDir)
doLast {
// NOTE: Use "public/" here so the export pass will find it and put it into the final site
genOutputDir.get().file("public/rss.xml").asFile.apply {
parentFile.mkdirs()
writeText(/* ... */)
println("Generated $absolutePath")
}
}
}
kotlin {
configAsKobwebApplication()
commonMain.dependencies { /* ... */ }
jsMain {
resources.srcDir(generateResourceTask) // <----- Set your task here
dependencies { /* ... */ }
}
}
Currently, Kobweb is still under active development, and due to our limited resources, we are focusing on improving the path to creating a new project from scratch. However, some users have shown interest in Kobweb but already have an existing project and aren't sure how to add Kobweb into it.
As long as you understand that this path isn't officially supported yet, we'll provide steps below to take which may
help people accomplish this manually for now. Honestly, the hardest part is creating a correct .kobweb/conf.yaml
,
which the following steps help you work around:
- Be sure to check the Kobweb compatibility matrix (see: COMPATIBILITY.md) to make sure you can match the versions it expects.
- Create a dummy app project somewhere. Pay attention to the questions it asks you, as you may want to choose a
package name that matches your project.
# In some tmp directory somewhere kobweb create app # or `kobweb create app/empty`, if you are already # experienced with Kobweb and know what you're doing
- When finished, copy the
site
subfolder out into your own project. (Once done, you can delete the dummy project, as it has served its usefulness.)cp -r app/site /path/to/your/project # delete app
- In your own project's root
settings.gradle.kts
file, include the new module and add our custom artifact repository link so your project can find the Kobweb Gradle plugins.// settings.gradle.kts pluginManagement { repositories { // ... other repositories you already declared ... maven("https://us-central1-maven.pkg.dev/varabyte-repos/public") } } // ... other includes you already declared include(":site")
- In your project's root
build.gradle.kts
file, add our custom artifact repository there as well (so your project can find Kobweb libraries)// build.gradle.kts subprojects { repositories { // ... other repositories you already declared ... maven("https://us-central1-maven.pkg.dev/varabyte-repos/public") } } // If you prefer, you can just declare this directly inside the // repositories block in site's `build.gradle.kts` file, but I // like declaring my maven repositories globally.
- Kobweb uses version catalogs for its dependencies. Add or update your version catalog under
gradle/libs.versions.toml
[versions] jetbrains-compose = "..." # replace with actual version, see COMPATIBILITY.md! kobweb = "..." # replace with actual version kotlin = "..." # replace with actual version [libraries] kobweb-api = { module = "com.varabyte.kobweb:kobweb-api", version.ref = "kobweb" } kobweb-core = { module = "com.varabyte.kobweb:kobweb-core ", version.ref = "kobweb" } kobweb-silk = { module = "com.varabyte.kobweb:kobweb-silk", version.ref = "kobweb" } kobwebx-markdown = { module = "com.varabyte.kobwebx:kobwebx-markdown", version.ref = "kobweb" } silk-icons-fa = { module = "com.varabyte.kobwebx:silk-icons-fa", version.ref = "kobweb" } [plugins] jetbrains-compose = { id = "org.jetbrains.compose", version.ref = "jetbrains-compose" } kobweb-application = { id = "com.varabyte.kobweb.application", version.ref = "kobweb" } kobwebx-markdown = { id = "com.varabyte.kobwebx.markdown", version.ref = "kobweb" } kotlin-multiplatform = { id = "org.jetbrains.kotlin.multiplatform", version.ref = "kotlin" }
If everything is working as expected, you should be able to run Kobweb within your project now:
# In /path/to/your/project
cd site
kobweb run
If you're still having issues, you may want to connect with us▼ for support (but understand that getting Kobweb added to complex existing projects may not be something we can currently prioritize).
While you can always export your site manually on your machine, you may want to automate this process. A common solution for this is a GitHub workflow.
For your convenience, we include a sample workflow below that exports your site and then uploads the results (which can be downloaded from a link shown in the workflow summary page):
# .github/workflows/export-site.yml
name: Export Kobweb site
on:
workflow_dispatch:
jobs:
export_and_upload:
runs-on: ubuntu-latest
defaults:
run:
shell: bash
env:
KOBWEB_CLI_VERSION: 0.9.18
steps:
- uses: actions/checkout@v4
- uses: actions/setup-java@v4
with:
distribution: temurin
java-version: 11
# When projects are created on Windows, the executable bit is sometimes lost. So set it back just in case.
- name: Ensure Gradle is executable
run: chmod +x gradlew
- name: Setup Gradle
uses: gradle/actions/setup-gradle@v3
- name: Query Browser Cache ID
id: browser-cache-id
run: echo "value=$(./gradlew -q :site:kobwebBrowserCacheId)" >> $GITHUB_OUTPUT
- name: Cache Browser Dependencies
uses: actions/cache@v4
id: playwright-cache
with:
path: ~/.cache/ms-playwright
key: ${{ runner.os }}-playwright-${{ steps.browser-cache-id.outputs.value }}
- name: Fetch kobweb
uses: robinraju/[email protected]
with:
repository: "varabyte/kobweb-cli"
tag: "v${{ env.KOBWEB_CLI_VERSION }}"
fileName: "kobweb-${{ env.KOBWEB_CLI_VERSION }}.zip"
tarBall: false
zipBall: false
- name: Unzip kobweb
run: unzip kobweb-${{ env.KOBWEB_CLI_VERSION }}.zip
- name: Run export
run: |
cd site
../kobweb-${{ env.KOBWEB_CLI_VERSION }}/bin/kobweb export --notty --layout static
- name: Upload site
uses: actions/upload-artifact@v4
with:
name: site
path: site/.kobweb/site/
if-no-files-found: error
retention-days: 1
You can copy this workflow (or parts of it) into your own GitHub project and then modify it to your needs.
Some notes...
- workflow_dispatch: This means that you can manually trigger this workflow from the GitHub UI, which I suggested here to prevent running an expensive export operation more than you need to. Of course, you can also configure your workflow to run on a schedule, or on push to a branch, etc.
- Setup Gradle: This action is optional but I recommend it because it configures a bunch of caching for you.
- Caching the browser:
kobweb export
needs to download a browser the first time it is run. This workflow sets up a cache that saves it across runs. The cache is tagged with a unique ID so that future Kobweb releases, which may change the version of the browser downloaded, will use a new cache bucket (allowing GitHub to eventually clean up the old one). - Upload site: This action uploads the exported site as an artifact. You can then download the artifact from the
workflow summary page. Your own workflow will likely delete this action and do something else here, like upload to a
web server (or some location accessible by your web server) or copy files over into a
gh_pages
repository. I've included this here (and set the retention days very low) just so you can verify that the workflow is working for your project.
For a simple site, the above workflow should take about 2 minutes to run.
StyleVariable
s work in a subtle way that is usually fine until it isn't -- which is often when you try to interact
with their values instead of just passing them around.
Specifically, this would compile but be a problem at runtime:
val MyOpacityVar by StyleVariable<Number>()
// later...
// Border opacity should be more opaque than the rest of the widget
val borderOpacity = max(1.0, MyOpacityVar.value().toDouble() * 2)
To see what the problem is, let's first take a step back. The following code:
val MyOpacityVar by StyleVariable<Number>()
// later...
Modifier.opacity(MyOpacityVar.value())
generates the following CSS:
opacity: var(--my-opacity);
However, MyOpacityVar
acts like a Number
in our code! How does something that effectively has a type of Number
generate text output like var(--my-opacity)
?
This is accomplished through the use of Kotlin/JS's unsafeCast
, where you can tell the compiler to treat a value as a
different type than it actually is. In this case, MyOpacityVar.value()
returns some object which the Kotlin compiler
treats like a Number
for compilation purposes, but it is really some class instance whose toString()
evaluates to
var(--my-opacity)
.
Therefore, Modifier.opacity(MyOpacityVar.value())
works seemingly like magic! However, if you try to do some
arithmetic, like MyOpacityVar.value().toDouble() * 0.5
, the compiler might be happy, but things will break silently at
runtime, when the JS engine is asked to do math on something that's not really a number.
In CSS, doing math with variables is accomplished by using calc
blocks, so Kobweb offers its own calc
method to
mirror this. When dealing with raw numerical values, you must wrap them in num
so we can escape the raw type system
which was causing runtime confusion above:
calc { num(MyOpacityVar.value()) * num(0.5) }
// Output: "calc(var(--my-opacity, 1) * 0.5)"
At this point, you can write code like this:
Modifier.opacity(calc { num(MyOpacityVar.value()) * num(0.5) })
It's a little hard to remember to wrap raw values in num
, but you will get compile errors if you do it wrong.
Working with variables representing length values don't require calc blocks because Compose HTML supports mathematical operations on such numeric unit types:
val MyFontSizeVar by StyleVariable<CSSLengthNumericValue>()
MyFontSizeVar.value() + 1.cssRem
// Output: "calc(var(--my-font-size) + 1rem)"
However, a calc block could still be useful if you were starting with a raw number that you wanted to convert to a size:
val MyFontSizeScaleFactorVar by StyleVariable<Number>()
calc { MyFontSizeScaleFactorVar.value() * 16.px }
// Output: calc(var(--my-font-size-scale-factor) * 16px)
Many users who create a full stack application generally expect to completely own both the client- and server-side code.
However, being an opinionated framework, Kobweb provides a custom Ktor server in order to deliver some of its features. For example, it implements the logic for handling server API routes▲ as well as some live reloading functionality.
It would not be trivial to refactor this behavior into some library that users could import into their own backend
server. As a compromise, some server configuration is exposed by the .kobweb/conf.yaml
file, and this has been the
main way users could affect the server's behavior.
That said, there will always be some use cases that Kobweb won't anticipate. So as an escape hatch, Kobweb allows users who know what they're doing to write their own plugins to extend the server.
Note
The Kobweb Server plugin feature is still fairly new. If you use it, please consider filing issues for any missing features and connecting with us▼ to share any feedback you have about your experience.
Creating a Kobweb server plugin is relatively straightforward. You'll need to:
- Create a new module in your project that produces a JAR file that bundles an implementation of
the
KobwebServerPlugin
interface. - Add that module as a
kobwebServerPlugin
dependency in your site's build script.- This ensures a copy of that jar is put under your project's
.kobweb/server/plugins
directory.
- This ensures a copy of that jar is put under your project's
The following steps will walk you through creating your first Kobweb Server Plugin.
Tip
You can
download this project to see the
completed result from applying the instructions in this section to the kobweb create app
site.
- Create a new module in your project.
- For example, name it "demo-server-plugin".
- Be sure to update your
settings.gradle.kts
file to include the new project.
- Add new entries for the
kobweb-server-project
library and kotlin JVM plugin in.gradle/libs.versions.toml
:[libraries] kobweb-server-plugin = { module = "com.varabyte.kobweb:kobweb-server-plugin", version.ref = "kobweb" } [plugins] kotlin-jvm = { id = "org.jetbrains.kotlin.jvm", version.ref = "kotlin" }
- For all remaining steps, create all files / directories under your new module's directory (e.g.
demo-server-plugin/
). - Create
build.gradle.kts
:plugins { alias(libs.plugins.kotlin.jvm) } group = "org.example.app" // update to your own project's group version = "1.0-SNAPSHOT" dependencies { compileOnly(libs.kobweb.server.plugin) }
- Create
src/main/kotlin/DemoKobwebServerPlugin.kt
:import com.varabyte.kobweb.server.plugin.KobwebServerPlugin import io.ktor.server.application.Application import io.ktor.server.application.log class DemoKobwebServerPlugin : KobwebServerPlugin { override fun configure(application: Application) { application.log.info("REPLACE ME WITH REAL CONFIGURATION") } }
Tip
As the Kobweb server is written in Ktor, you should familiarize yourself with Ktor's documentation.
- Create
src/main/resources/META-INF/services/com.varabyte.kobweb.server.plugin.KobwebServerPlugin
, setting its content to the fully-qualified class name of your plugin. For example:org.example.app.DemoKobwebServerPlugin
Note
If you aren't familiar with META-INF/services
, you can
read this helpful article to learn more about service implementations, a very
useful Java feature.
The Kobweb Gradle Application plugin provides a way to notify it about your JAR project. Set it up, and Gradle will build and copy your plugin jar over for you automatically.
In your Kobweb project's build script, include the following kobwebServerPlugin
line in a top-level dependencies
block:
// site/build.gradle.kts
dependencies {
kobwebServerPlugin(project(":demo-server-plugin"))
}
kotlin { /* ... */ }
Important
You need to put the kobwebServerPlugin
declaration inside a top-level dependencies
block, not in one of the
ones nested under the kotlin
block (such as kotlin.jvmMain.dependencies
).
Once this is set up, upon the next Kobweb server run (e.g. via kobweb run
), if you check the logs, you should see
something like this:
[main] INFO ktor.application - Autoreload is disabled because the development mode is off.
[main] INFO ktor.application - REPLACE ME WITH REAL CONFIGURATION
[main] INFO ktor.application - Application started in 0.112 seconds.
[main] INFO ktor.application - Responding at http://0.0.0.0:8080
Despite the simplicity of the KobwebServerPlugin
interface, the application
parameter passed
into KobwebServerPlugin.configure
is quite powerful.
While I know it may sound kind of meta, you can create and install a Ktor Application Plugin inside a Kobweb Server Plugin. Once you've done that, you have access to all stages of a network call, as well as some other hooks like ones for receiving Application lifecycle events.
Tip
Please read the Extending Ktor documentation to learn more.
Doing so looks like this:
import com.varabyte.kobweb.server.plugin.KobwebServerPlugin
import io.ktor.server.application.Application
import io.ktor.server.application.createApplicationPlugin
import io.ktor.server.application.install
class DemoKobwebServerPlugin : KobwebServerPlugin {
override fun configure(application: Application) {
val demo = createApplicationPlugin("DemoKobwebServerPlugin") {
onCall { call -> /* ... */ } // Request comes in
onCallRespond { call -> /* ... */ } // Response goes out
}
application.install(demo)
}
}
It's important to note that, unlike other parts of Kobweb, Kobweb Server Plugins do NOT support live reloading. We only start up and configure a Kobweb server once in its lifetime.
If you make a change to a Kobweb Server Plugin, you must quit and restart the server for it to take effect.
You may already have an existing and complex backend, perhaps written with Ktor or Spring Boot, and, if so, are wondering if you can integrate Kobweb with it.
The recommended solution for now is to export your site using a static layout (read more about static layout sites here▲) and then add code to your backend to serve the files yourself, as it is fairly trivial.
When you export a site statically, it will generate all files into your .kobweb/site
folder. Then, if using Ktor, for
example, serving these files is a one-liner:
routing {
staticFiles("/", File(".kobweb/site"))
}
If using Ktor, you should also install
the IgnoreTrailingSlash
plugin
so that your web server will serve index.html
when a user visits a directory (e.g. /docs/
) instead of returning a 404:
embeddedServer(...) { // `this` is `Application` in this scope
this.install(IgnoreTrailingSlash)
// Remaining configuration
}
If you need to access HTTP endpoints exposed by your backend, you can use window.fetch(...)
directly, or you can use the convenience http
property that Kobweb adds to the window
object which exposes
all the HTTP methods (get
, post
, put
, etc.):
@Page
@Composable
fun CustomBackendDemoPage() {
LaunchedEffect(Unit) {
val endpointResponse = window.http.get("/my/endpoint?id=123").decodeToString()
/* ... */
}
}
Unfortunately, using your own backend does mean you're opting out of using Kobweb's full stack solution, which means you won't have access to Kobweb's API routes, API streams, or live reloading support. This is a situation we'd like to improve someday (link to tracking issue), but we don't have enough resources to be able to prioritize resolving this for a 1.0 release.
Kobweb introduces a handful of type-aliases for CSS unit values, basing them off of the CSSNumericValue
class and
extending the set defined by Compose HTML:
typealias CSSAngleNumericValue = CSSNumericValue<out CSSUnitAngle>
typealias CSSLengthOrPercentageNumericValue = CSSNumericValue<out CSSUnitLengthOrPercentage>
typealias CSSLengthNumericValue = CSSNumericValue<out CSSUnitLength>
typealias CSSPercentageNumericValue = CSSNumericValue<out CSSUnitPercentage>
typealias CSSFlexNumericValue = CSSNumericValue<out CSSUnitFlex>
typealias CSSTimeNumericValue = CSSNumericValue<out CSSUnitTime>
This section explains why they were added and why you should almost always prefer using them.
When you write CSS values like 10.px
, 5.cssRem
, 45.deg
, or even 30.s
into your code, you normally don't have to
think too much about their types. You just create them and pass them into the appropriate Kobweb / Compose HTML APIs.
Let's discuss what is actually happening when you do this. Compose HTML provides a CSSSizeValue
class which represents
a number value and its unit.
val lengthValue = 10.px // CSSSizeValue<CSSUnit.px> (value = 10 and unit = px)
val angleValue = 45.deg // CSSSizeValue<CSSUnit.deg> (value = 45 and unit = deg)
This is a pretty elegant approach, but the types are verbose. This can be troublesome when writing code that needs to work with them:
val lengths: List<CSSSizeValue<CSSUnit.px>>
fun drawArc(arc: CSSSizeValue<CSSUnit.deg>)
Note also that the above cases are overly restrictive, only supporting a single length and angle type, respectively. We
usually want to support all relevant types (e.g. px
, em
, cssRem
, etc. for lengths; deg
, rad
, grad
, and
turn
for angles). We can do this with the following out
syntax:
val lengths: List<CSSSizeValue<out CSSUnitLength>>
fun drawArc(arc: CSSSizeValue<out CSSUnitAngle>)
What a mouthful!
As a result, the Compose HTML team added type-aliases for all these unit types, such as CSSLengthValue
and CSSAngleValue
. Now, you can write the above code like:
val lengths: List<CSSLengthValue>
fun drawArc(arc: CSSAngleValue)
Much better! Seems great. No problems, right? Right?!
You can probably tell by my tone: Yes problems.
To explain, we first need to talk about CSSNumericValue
.
It is common to transform values in CSS using many of its various mathematical functions. Perhaps you want to take the
sum of two different units (10.px + 5.cssRem
) or call some other math function (clamp(1.cssRem, 3.vw)
). These
operations return intermediate values that cannot be directly queried like a CSSSizeValue
can.
This is handled by the CSSNumericValue
class, also defined by Compose HTML (and which is actually a base class
of CSSSizeValue
).
val lengthSum = 10.px + 2.cssRem // CSSNumericValue<CSSUnitLength>
val angleSum = 45.deg + 1.turn // CSSNumericValue<CSSAngleLength>
These numeric operations are of course useful to the browser, which can resolve them into absolute screen values, but for us in user space, they are opaque calculations.
In practice, however, that's fine! The limited view of these values does not matter because we rarely need to query them in our code. In almost all cases, we just take some numeric value, optionally tweak it by doing some more math on it, and then pass it onto the browser.
Because it is opaque, CSSNumericValue
is far more flexible and widely applicable than CSSSizeValue
is. If you are
writing a function that takes a parameter, or declaring a StyleVariable
tied to some length or time, you almost always
want to use CSSNumericValue
and not CSSSizeValue
.
As mentioned above, the Compose HTML team created their unit-related type-aliases against the CSSSizeValue
class.
This decision makes it really easy to write code that works well when you test it with concrete size values but is actually more restrictive than you expected.
Kobweb ensures its APIs all reference its CSSNumericValue
type-aliases:
// Legacy Kobweb
fun Modifier.lineHeight(value: CSSLengthOrPercentageValue): Modifier = styleModifier {
lineHeight(value)
}
// Modern Kobweb
fun Modifier.lineHeight(value: CSSLengthOrPercentageNumericValue): Modifier = styleModifier {
lineHeight(value)
}
If you are using style variables in your code, or writing your own functions that take CSS units as arguments, you might be referencing the Compose HTML types. Your code will still work fine, but you are strongly encouraged to migrate them to Kobweb's newer set, in order to make your code more flexible about what it can accept:
// Not recommended
val MyFontSize by StyleVariable<CSSLengthValue>
fun drawArc(arc: CSSAngleValue)
// Recommended
val MyFontSize by StyleVariable<CSSLengthNumericValue>
fun drawArc(arc: CSSAngleNumericValue)
Note
Perhaps in the future, the Compose HTML team might consider updating their type-aliases to use the CSSNumericValue
type and not the CSSSizeValue
type. If that happens, we can revert our changes and delete this section. But until
then, it's worth understanding why Kobweb introduces its own type-aliases and why you are encouraged to use them
instead of the Compose HTML versions.
A Kobweb project always has a frontend and, if configured as a full stack site, a backend as well. Both require different steps to debug them.
At the moment, attaching a debugger to Kotlin/JS code requires IntelliJ Ultimate. If you have it, you can follow these steps in the official docs.
Important
Be sure the port in your URL matches the port you specified in your .kobweb/conf.yaml
file. By default, this
is 8080.
If you do not have access to IntelliJ Ultimate, then you'll have to rely on println
debugging. While this is far from
great, live reloading plus Kotlin's type system generally help you incrementally build your site up without too many
issues.
Tip
If you're a student, you can apply for a free IntelliJ Ultimate license here. If you maintain an open source project, you can apply here.
Debugging the backend first requires configuring the Kobweb server to support remote debugging. This is easy to do by
modifying the kobweb
block in your build script to enable remote debugging:
kobweb {
app {
server {
remoteDebugging {
enabled.set(true)
port.set(5005)
}
}
}
}
Note
Specifying the port is optional. Otherwise, it is 5005, a common remote debugging default. If you ever need to debug multiple Kobweb servers at the same time, however, it can be useful to change it.
Once you've enabled remote debugging support, you can then follow the official documentation to add a remote JVM debug configuration to your IDE.
Important
For remote debugging to work:
- The Debugger Mode should be set to Attach to remote JVM.
- You need to correctly specify the Use module classpath value. In general, use the
jvmMain
classpath associated with your Kobweb application, e.g.app.site.jvmMain
. If you've refactored your backend code out to another module, you should be able to use that instead.
At this point, start up your Kobweb server using kobweb run
.
Caution
Remote debugging is only supported in dev mode. It will not be enabled for a server started with
kobweb run --env prod
.
With your Kobweb server running and your "remote debug" run configuration selected, press the debug button. If
everything is set up correctly, you should see a message in the IDE debugger console
like: Connected to the target VM, address: 'localhost:5005', transport: 'socket'
If instead, you see a red popup with a message
like Unable to open debugger port (localhost:5005): java.net.ConnectException "Connection refused"
, please
double-check the values in your conf.yaml
file, restart the server, and try again.
The easiest way to use a custom font is if it is already hosted for you. For example, Google Fonts provides a CDN that you can use to load fonts directly.
Caution
While this is the easiest approach, be sure you won't run into compliance issues! If you use Google Fonts on your site, you may technically be in violation of the GDPR in Europe, because an EU citizen's IP address is communicated to Google and logged. You may wish to find a Europe-safe host instead, or self-host, which you can read about in the next section▼.
The font service should give you HTML to add to your site's <head>
tag. For example, Google Fonts suggests the
following when I select Roboto Regular 400:
<link rel="preconnect" href="https://fonts.googleapis.com">
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
<link href="https://fonts.googleapis.com/css2?family=Roboto&display=swap" rel="stylesheet">
This code should be converted into Kotlin and added to the kobweb
block of your site's build.gradle.kts
script:
kobweb {
app {
index {
head.add {
link(rel = "preconnect", href = "https://fonts.googleapis.com")
link(rel = "preconnect", href = "https://fonts.gstatic.com") { attributes["crossorigin"] = "" }
link(
href = "https://fonts.googleapis.com/css2?family=Roboto&display=swap",
rel = "stylesheet"
)
}
}
}
}
Once done, you can now reference this new font:
Column(Modifier.fontFamily("Roboto")) {
Text("Hello world!")
}
Users can flexibly declare a custom font by using
CSS's @font-face
rule.
In Kobweb, you can normally declare CSS properties in Kotlin (within an @InitSilk
block), but unfortunately, Firefox
doesn't allow you to define or modify @font-face
entries
in code (relevant Bugzilla issue). Therefore, for guaranteed
cross-platform compatibility, you should create a CSS file and reference it from your build script.
To keep the example concrete, let's say you've downloaded the open source font Lobster from Google Fonts (and its license as well, of course).
You need to put the font file inside your public resources directory, so it can be found by the user visiting your site. I recommend the following file organization:
jsMain
└── resources
└── public
└── fonts
├── faces.css
└── lobster
├── OFL.txt
└── Lobster-Regular.ttf
where faces.css
contains all your @font-face
rule definitions (we just have a single one for now):
@font-face {
font-family: 'Lobster';
src: url('/fonts/lobster/Lobster-Regular.ttf');
}
Note
The above layout may be slightly overkill if you are sure you'll only ever have a single font, but it's flexible enough to support additional fonts if you decide to add more in the future, which is why we recommend it as a general advice here.
Now, you need to reference this CSS file from your build.gradle.kts
script:
kobweb {
app {
index {
head.add {
link(rel = "stylesheet", href = "/fonts/faces.css")
}
}
}
}
Finally, you can reference the font in your code:
Column(Modifier.fontFamily("Lobster")) {
Text("Hello world!")
}
When you run kobweb run
, the spun-up web server will, by default, log to the .kobweb/server/logs
directory.
Note
You can generate logs using the ctx.logger
property inside @Api
calls▲.
You can configure logging behavior by editing the .kobweb/conf.yaml
file. Below we show setting all parameters to
their default values:
server:
logging:
level: DEBUG # ALL, TRACE, DEBUG, INFO, WARN, ERROR, OFF
logRoot: ".kobweb/server/logs"
clearLogsOnStart: true # Warning - if true, wipes ALL files in logRoot, so don't put other files in there!
logFileBaseName: "kobweb-server" # e.g. "kobweb-server.log", "kobweb-server.2023-04-13.log"
maxFileCount: null # null = unbound. One log file is created per day, so 30 = 1 month of logs
totalSizeCap: 10MiB # null = unbound. Accepted units: B, K, M, G, KB, MB, GB, KiB, MiB, GiB
compressHistory: true # If true, old log files are compressed with gzip
The above defaults were chosen to be reasonable for most users running their projects on their local machines in
developer mode. However, for production servers, you may want to set clearLogsOnStart
to false, bump up the
totalSizeCap
after reviewing the disk limitations of your web server host, and maybe set maxFileCount
to a reasonable
limit.
Note that most config files assume "10MB" is 10 * 1024 * 1024 bytes, but here it will actually result in 10 * 1000 * 1000 bytes. You probably want to use "KiB", "MiB", or "GiB" when you configure this value.
CORS, or Cross-Origin Resource Sharing, is a security feature built on the idea that a web page should not be able to make requests for resources from a server that is not the same as the one that served the page unless it was served from a trusted domain.
To configure CORS for a Kobweb backend, Kobweb's .kobweb/conf.yaml
file allows you to declare such trusted domains
using a cors
block:
server:
cors:
hosts:
- name: "example.com"
schemes:
- "https"
Note
Specifying the schemes is optional. If you don't specify them, Kobweb defaults to "http" and "https".
Note
You can also specify subdomains, e.g.
- name: "example.com"
subdomains:
- "en"
- "de"
- "es"
which would add CORS support for en.example.com
, de.example.com
, and es.example.com
, as well as example.com
itself.
Once configured, your Kobweb server will be able to respond to data requests from any of the specified hosts.
Tip
If you find that your full-stack site, which was working locally during development, rejects requests in the production version, check your browser's console logs. If you see errors in there about a violated CORS policy, that means you didn't configure CORS correctly.
The Kobweb export feature is built on top of Microsoft Playwright, a solution for making it easy to download and run browsers programmatically.
One of the features provided by Playwright is the ability to generate traces, which are essentially detailed reports
you can use to understand what is happening as your site loads. Kobweb exposes this feature through the export
block
in your Kobweb application's build script.
Enabling traces is easy:
// build.gradle.kts
plugins {
// ... other plugins ...
alias(libs.plugins.kobweb.application)
}
kobweb {
app {
export {
enableTraces()
}
}
}
You can pass in parameters to configure the enableTraces
method, but by default, it will generate trace files into
your .kobweb/export-traces/
directory.
Once enabled, you can run kobweb export
, then once exported, open any of the generated *.trace.zip
files by
navigating to them using your OS's file explorer and drag-and-dropping them into
the Playwright Trace Viewer.
Tip
You can learn more about how to use the Trace Viewer using the official documentation.
It's not expected many users will need to debug their site exports, but it's a great tool to have (especially combined with the server logs feature) to diagnose if one of your pages is taking longer to export than expected.
In the beginning, Kobweb was only intended to be a thin layer on top of Compose HTML, but the more we worked on it, the more we ran into features that were simply not yet implemented in Compose HTML. In other cases, we found ourselves reaching for utilities that we wished existed in Kotlin/JS browser APIs. As we began adding these features, we realized it would have been a shame to bury them deep inside our framework.
As a result, we created two modules:
compose-html-ext
, where we put code that we would be more than happy for the Compose HTML team to fork and migrate over to Compose HTML someday.browser-ext
, a collection of general purpose utilities that we think could be useful to any Kotlin/JS project targeting the browser.
The features across these modules include (not comprehensive):
- a ton of missing type-safe wrappers around many, many CSS properties
- type-safe wrappers around CSS functions, like gradients, filters,
calc
(especially useful when working with CSS variables), etc. - rich SVG support
- utility methods around saving files to / loading files from the disk
- utility methods and classes built on top
of
window.fetch
(for example, making it easier to use the most common HTTP verbs like GET, POST, etc., as well as providingsuspend fun
versions of fetch) - additions for the missing transition events
- implementations of resize and intersection observers
- utility methods for getting a sequence of descendant/ancestor HTML elements, useful for walking the DOM tree in a Kotlin-idiomatic way
- a utility composable,
GenericTag
, which is an easy-to-use API wrapping Compose HTML'sTagElement
composable, with additional namespacing support if needed (for example, required when implementing SVG elements) - a utility class for working with CSS variables,
StyleVariable
, allows specifying a default value, provides first-class number/string variable support, and fixes a bug in Compose HTML'sCSSStyleVariable
class where it can accept invalid values. setTimeout
andsetInterval
methods that are more Kotlin-idiomatic (e.g. the lambdas are the last parameter)
Note
Some users have mentioned we should have opened PRs for the Compose HTML team instead of maintaining a separate codebase. However, after observing that JetBrains was focusing more and more of its energy on Compose Multiplatform for Web, we decided to implement the features we needed in our own project. This way, we could maintain our velocity while allowing their team to pick and choose what they agreed with at some point in the future at their leisure. There's so much code here, especially around CSS APIs, that getting mired down in PR discussions would have ground our progress to a halt.
If you want to use Compose HTML but not Kobweb, or Kotlin/JS but not Compose HTML, you can still use and benefit
from compose-html-ext
or browser-ext
in your own project. An example build script could look like this (here, for a
non-Kobweb Compose HTML project):
// build.gradle.kts
plugins {
kotlin("multiplatform") version "..."
}
repositories {
mavenCentral()
maven("https://maven.pkg.jetbrains.space/public/p/compose/dev")
google()
maven("https://us-central1-maven.pkg.dev/varabyte-repos/public") // IMPORTANT!!!
}
kotlin {
js().browser()
sourceSets {
jsMain.dependencies {
implementation(compose.html.core)
implementation(compose.runtime)
implementation("com.varabyte.kobweb:compose-html-ext:...") // IMPORTANT!!!
}
}
}
Note
The compose-html-ext
dependency automatically exposes the browser-ext
dependency.
Jetbrains is working on the "Compose Multiplatform UI Framework", which allows developers to use the same codebase across Android, iOS, Desktop, and the Web. And it may seem like the Kobweb + Silk approach is obsoleted by it.
It's first worth understanding the core difference between the two approaches. With Compose Multiplatform, the framework owns its own rendering pipeline, drawing to a buffer. In contrast, Compose HTML modifies an HTML / CSS DOM tree and leaves it up to the browser to do the final rendering.
This has major implications on how similar the two APIs can get. For example, in Compose Multiplatform, the order you apply modifiers matters. However, in Compose HTML, this action simply sets html style properties under the hood, where order does not matter.
Due to its reputation, ditching HTML / CSS entirely at first can seem like a total win, but this approach has several limitations:
- robots would lose the ability to crawl and index your site, hurting SEO.
- your initial render may take longer (as nothing will be rendered until your site's logic is downloaded and run for one frame).
- your site will need to allocate a large canvas buffer, which could be very expensive on high-res, wide-screen desktops.
- your UI will be opaque to the powerful suite of devtools that come bundled with browsers.
- you won't have the ability to style unvisited vs visited links differently (this information is hidden from you by the browser for security reasons and can only be set through HTML / CSS).
- you won't have the ability to turn elements on / off when printing the page.
- accessibility tools for browsers might not work.
- the download size of the rendering components is not insignificant and apparently not very compressible, often resulting in a site's basic footprint being 4-6x larger total (e.g. 200-400K vs. 2-3MB for small sites).
It would also prevent a developer from making use of the rich ecosystem of Javascript libraries out there.
Finally, Kobweb is more than just Kotlin-ifying HTML / CSS. It also provides rich integration with powerful web technologies like web workers▲ and websockets▲.
For now, I am making a bet that there will always be value in embracing the web, providing a framework that sticks to HTML / CSS but offers a growing suite of UI widgets, layouts, and other features that make it a more comfortable experience for the Kotlin developer.
For example, the flexbox layout is a very powerful concept,
but it can be very tricky to use. In most cases, you'll find it's much easier to compose Row
s and Column
s together
than trying to remember if you should be justifying your items or aligning your content, even if Row
s and Column
s
are just configuring the correct HTML / CSS for you behind the scenes.
Ultimately, I believe there is room for both Compose Multiplatform and Kobweb. If you want to make an app experience that feels the same on Android, iOS, Desktop, and Web, then Compose Multiplatform could be the right choice for you. However, if you just want to make a traditional website but want to use Kotlin instead of TypeScript, Kobweb can provide an excellent development experience for that case.
Current state: Foundations are in place! You may encounter API gaps.
You may wish to refer to our Kobweb 1.0 roadmap document.
Kobweb is becoming quite functional. We are already using it to build https://kobweb.varabyte.com and https://bitspittle.dev. Several users have created working portfolio sites already, and I'm aware of at least two cases where Kobweb was used in a project for a client.
At this point:
- It is easy to set up a new project and get things running quickly.
- The live reloading flow is pretty nice, and you'll miss it when you switch to projects that don't have it.
- It supports generating pages from Markdown that can reference your Composable code.
- While it's not quite a server-side rendering, you can export static pages which will get hydrated on load.
- A huge range of CSS properties are supported, along with support for style variables and animations.
- You can use the
Modifier
builder for a significant number of CSS properties. - Silk components are color-mode aware and support responsive behavior.
- There are quite a few widgets available, and it's easy to create your own.
However, there's always more to do.
- I'm trying to add support for every stabilized CSS property, but some are still missing, especially less common ones. (You can use a fallback for such cases in the meantime).
- There are still a handful of widgets planned to be added.
- A lot of detailed documentation is planned to go into the Kobweb site (linked just above) but it isn't done yet.
I think there's enough there now to let you do almost anything you'd want to do, as either Kobweb supports it or you can escape hatch to underlying Compose HTML / Kotlin/JS approaches, but there might be some areas where it's still a bit DIY. It would be great to get real-world experience to hear what issues users are actually running into.
So, should you use Kobweb at this point? If you are...
- playing around with Compose HTML for the first time and want to get up and running quickly on a toy project:
- YES!!! Please see the connecting with us▼ section below, we'd definitely love to hear from you. It's still a good time if you want to have a voice in the direction of this project.
- a Kotlin developer who wants to write a small web app or create a new blog from scratch:
- Probably! I hope if you evaluate Kobweb at this point, you'll find a lot to like. You can get in touch with us at our Discord if you try it and have questions or run into missing features.
- someone who already has an existing project in progress and wants to integrate Kobweb into it:
- Maybe not? Depending on how much work you've done, it may not be a trivial refactor. You can review this earlier section▲ if you want to try anyway.
- a company:
- Probably not? I'm assuming most companies are so risk-averse they would not even use Compose HTML, which Kobweb is built on top of. If you were considering Compose HTML, however, Kobweb is worth a look.
On the fence but not sure? Connect with us, and I'd be happy to help you assess your situation.
I'm pleased to mention that Kobweb has received feedback from some satisfied users. Here are a few:
- "This is a pretty bloody amazing technology you've created here. I have been dreading upgrading [my] website for ages because I didn't want to go back to html and css 🫤 now I can stay with Kotlin 😀"
- "Kobweb looks fantastic and I've been [trying] to use Kotlin in all parts of [my] hobby stuff and work, so I got real excited when I saw Kobweb, [even though] I hadn't been satisfied with a web framework in a long time. Incredible work."
- "I started using Kobweb last week and I have to say this [...] reinvented web development for me. [...] I used to hate html css. After getting my hands on kobweb I’m in love with it."
- "Finally got paid -- all thanks to kobweb 🎉💥"
- "I didn't wanna learn any JS framework so when I first learned about kobweb it felt like a no-brainer; having built 2 Android apps with compose already and a backend with ktor. One could argue Android developers are the best target audience since the additional knowledge needed to move an app to the web with Kobweb is minimal. I love it! 🤩"
Kobweb uses BrowserStack when we need to test its APIs on older browsers.
We appreciate their support for the open source community.
- Join my Discord!
- GitHub Discussions for this project
- The Kobweb channel on the Kotlin Slack
- You can send direct queries to my email
If you're comfortable with it, using Discord is recommended, because there's a growing community of users in there who can offer help even when I'm not around.
It is still early days, and while we believe we've proven the feasibility of this approach at this point, there's still plenty of work to do to get to a 1.0 launch! We are hungry for the community's feedback, so please don't hesitate to:
- Open an issue
- Contact us (using any of the ways mentioned above) telling us what features you want
- Ask us for guidance, especially as there are no tutorials yet (your questions can help us know what to write first!)
Thank you for your support and interest in Kobweb!