JavaScript platform heavily uses asyncronous actions that requires continuation callbacks. Codes using them easily become complex for nested callbacks.
One of profit of using promises is changing the callback nesting style to more clean.
The promises may be integrated to ES6 (harmony). Q is an implementation of the promises.
Q is a module for promises in JavaScript.
In node.js or JS on HTML, using asynchronous actions is usually coded as the function callback style. e.g.:
// using-callback.js
http.get(url.parse("http://example.org/"), function (res) {
console.log(res.statusCode); // maybe 302
});
If you want to chain contimuous asynchronous actions, following callbacks are nested in the callback of the former action; e.g.:
// using-nested-callback.js
http.get(url.parse("http://example.org/"), function (res) {
console.log(res.statusCode); // maybe 302
http.get(url.parse(res.headers["location"]), function (res) {
console.log(res.statusCode); // maybe 200
});
});
Using promises with Q
transforms
the nested style to flatten manner, as:
// using-promise.js
httpGet(url.parse("http://example.org/")).then(function (res) {
console.log(res.statusCode); // maybe 302
return httpGet(url.parse(res.headers["location"]));
}).then(function (res) {
console.log(res.statusCode); // maybe 200
});
From the flatten style, promises introduce many ways of clearer coding to JavaScript programmings.
At last, the promised action httpGet
can be just defined as:
// using-promise.js
var httpGet = function (opts) {
var deferred = Q.defer();
http.get(opts, deferred.resolve);
return deferred.promise;
};
Making promise actions is also transforming from callback style actions.
If you use in node.js, simply install with npm:
npm install q
then using in js file as an usual module:
var Q = require("q");
...
If you use in browsers, loading q.min.js
in former of your JavaScript code:
<!-- using-q.html -->
<script src="q.min.js"></script>
<script>
var loadText = function (url) {
var deferred = Q.defer();
var req = new XMLHttpRequest();
req.onreadystatechange = function () {
if (req.readyState !== 4) return;
if (/^[^2]\d\d$/.exec(req.status)) return deferred.reject(req.status);
deferred.resolve(req.responseText);
};
req.open("GET", url, true);
req.send();
return deferred.promise;
};
</script>
q.min.js
is in
the source repository of q.
At first using Q
, you make actions return "promise".
For example, make node.js action http.get
as promised action:
// using-promise.js
var httpGet = function (opts) {
var deferred = Q.defer();
http.get(opts, deferred.resolve);
return deferred.promise;
};
then you can use as:
httpGet(...).then(function (res) {...});
It is important that functions returns promise.
The first Q.defer()
returns set of an empty promise and operations for it.
deferred.promise
is the empty promise and deferred.resolve(value)
makes the promise resolved as the value
.
The simplest example of them is:
// promise-resolve-then-flow.js
var deferred = Q.defer();
deferred.promise.then(function (obj) {
console.log(obj);
});
deferred.resolve("Hello World");
It prints "Hello World"
to the console.
In generally, you can transform usual callback actions:
// promise-translate-action.js
action(arg1, arg2, function (result) {
doSomething(result);
});
to promise actions:
// promise-translate-action.js
var promiseAction = function (arg1, arg2) {
var deferred = Q.defer();
action(arg1, arg2, deferred.resolve);
return deferred.promise;
}
promiseAction(arg1, arg2).then(function (result) {
doSomething(result);
});
Using continuous actions is call next action in the former callback. e.g.:
// chain-nested-callback.js
http.get(url.parse("http://example.org"), function (res) {
http.get(url.parse(res.headers["location"]), function (res) {
var body = "";
res.on("data", function (chunk) {
body += chunk;
});
res.on("end", function () {
console.log(body);
});
});
});
You can simply transform it to promise actions and then
callback style:
// chain-nested-promise.js
httpGet(url.parse("http://example.org")).then(function (res) {
httpGet(url.parse(res.headers["location"])).then(function (res) {
var body = "";
res.on("data", function (chunk) {
body += chunk;
});
res.on("end", function () {
console.log(body);
});
});
});
If you can "return" any value in then
callback, you can catch the value to
a nextthen
callback:
// chain-by-value.js
httpGet(url.parse("http://example.org")).then(function (res) {
return res.headers["location"];
}).then(function (location) {
httpGet(url.parse(location)).then(function (res) {
var body = "";
res.on("data", function (chunk) {
body += chunk;
});
res.on("end", function () {
console.log(body);
});
});
});
If you return a "promise" in then
callback,
you can catch a resolved value of the promise to the next then
callback:
// chain-by-promise.js
httpGet(url.parse("http://example.org")).then(function (res) {
return httpGet(url.parse(res.headers["location"]));
}).then(function (res) {
var body = "";
res.on("data", function (chunk) {
body += chunk;
});
res.on("end", function () {
console.log(body);
});
});
You can transforn the body
loading action to promise action as:
// chain-to-action.js
var loadBody = function (res) {
var deferred = Q.defer();
var body = "";
res.on("data", function (chunk) {
body += chunk;
});
res.on("end", function () {
deferred.resolve(body);
});
return deferred.promise;
};
Apply it:
// chain-to-action.js
httpGet(url.parse("http://example.org")).then(function (res) {
return httpGet(url.parse(res.headers["location"]));
}).then(function (res) {
loadBody(res).then(function (body) {
console.log(body);
});
});
It can also go outside of then
callback as:
// chain-to-action-flatten.js
httpGet(url.parse("http://example.org")).then(function (res) {
return httpGet(url.parse(res.headers["location"]));
}).then(function (res) {
return loadBody(res);
}).then(function (body) {
console.log(body);
});
Make it more simple as:
// chain-to-action-simplify.js
httpGet(url.parse("http://example.org")).then(function (res) {
return httpGet(url.parse(res.headers["location"]));
}).then(loadBody).then(console.log);
You can use if
branch in promise as:
// chain-branch.js
var getBody = function (uri) {
return httpGet(url.parse(uri)).then(function (res) {
if (res.statusCode.toString()[0] === "3") {
return httpGet(url.parse(res.headers["location"]));
} else {
return res;
}
}).then(loadBody);
};
getBody("http://example.org").then(console.log);
The first returns promise of res
(http.ClientResponse)
that returns res
.
The second returns just res
.
every return
of branch should finally resolve same type object.
Branch in promise action is with this manner.
You can use loop in promise with recursive call:
// chain-loop.js
var redirectGet = function (uri) {
return httpGet(url.parse(uri)).then(function (res) {
if (res.statusCode.toString()[0] === "3") {
return redirectGet(res.headers["location"]);
} else {
return res;
}
});
};
var getBody = function (uri) {
return redirectGet(uri).then(loadBody);
};
getBody("http://example.org").then(console.log);
Q includes standard error handling ways.
The then
method accept two callback args;
the first callback catches successful result,
and the second catches error as failure.
// promise-failure-action.js
promisedAction(arg1, arg2).then(function (value) {
doSomething(value);
}, function (error) {
doErrorHandle(error);
});
Q.defer()
set has reject(error)
function for passing error handler.
A simple example is:
// promise-reject-then-flow.js
var deferred = Q.defer();
deferred.promise.then(null, console.log);
deferred.reject(new Error("failure"));
It prints [Error: failure]
.
If you can pass only error handing callback, use 'fail(handler)':
// promise-reject-fail-flow.js
var deferred = Q.defer();
deferred.promise.fail(console.log);
deferred.reject(new Error("failure"));
The real example httpGet
with error handling becomes as:
// using-promise-with-error.js
var httpGet = function (opts) {
var deferred = Q.defer();
try {
http.get(opts, deferred.resolve).on("error", deferred.reject);
} catch (error) {deferred.reject(error);}
return deferred.promise;
};
In a promise chain, when error occured in middle promise action, following actions not executed, the error is passed through to the latter promises that has an error handler. If all latter promises have not error handler, error is ignored.
For example:
// chain-with-error.js
httpGet(url.parse("http://example.biz")).then(function (res) {
return httpGet(url.parse(res.headers["location"]));
}).then(loadBody).then(console.log).fail(function (error) {
console.log("error occured: " + error);
});
The last error handler catches an error occured at any point in
httpGet
, loadBody
, or console.log
of then
.
An error thrown in the then
callback also becomes failure flow. e.g.:
// promise-throw-fail-flow.js
var deferred = Q.defer();
deferred.promise.then(function (obj) {
throw new Error("thrown error");
}).fail(console.log);
deferred.resolve({});
It prints [Error: thrown error]
.
Notice that a thrown error can be catched in the "following" promise.
So, the code:
// promise-throw-uncaught-flow.js
var deferred = Q.defer();
deferred.promise.then(function (obj) {
throw new Error("thrown error");
}, console.log);
deferred.resolve({});
It prints nothing.
The error flow is also a part of promise. Retrieving from error is just return a value in error callback. as:
// chain-with-error-recover.js
httpGet(url.parse("http://example.biz")).then(function (res) {
return httpGet(url.parse(res.headers["location"]));
}).then(loadBody).fail(function (error) {
return "Not Found";
}).then(console.log);
If error occured in httpGet
or loadBody
,
it prints "Not Found" as successful flow.
The then
callback always accepts just one argument.
So you want to pass multiple values as a array like [value1, value2]
.
In callback, extract with array access like: var value1 = result[0];
.
If you want to use multiple arguments directly,
you can use spread
instead of then
:
// promise-multiple-resolve-spread-flow.js
var deferred = Q.defer();
deferred.promise.spread(function (value1, value2) {
return value1 + value2;
}).then(console.log);
deferred.resolve([10, 30]);
Not only async action, usual value transform as a promise.
Q.resolve
transforms value to promise:
// promise-resolve.js
Q.resolve("Hello").then(console.log);
Not only a returned value of then
callback,
other types of value is handled with same manner of promises.
Q.when(value, success, failure)
is same as
Q.resolve(value).then(success, failure)
(but then
is implemented with when
in really)
// promise-when.js
Q.when("Hello", console.log);
Q.reject(obj)
is "rejected" promise.
it spawn only error handler. e.g.:
// promise-reject.js
Q.reject("abc").then(function (value) {
console.log("OK");
}, function (error) {
console.log("FAIL");
});
It just prints "FAIL"
.
If you return rejected promises in then
callback,
It spawn error handler. e.g.:
// promise-return-reject-flow.js
Q.resolve("abc").then(function (value) {
return Q.reject("FAIL");
}).then(null, function (error) {
console.log(error);
});
It also prints "FAIL"
.
Returning a reject promise is same effect of throw error
on flows.
The then
callback can use the result anyway.
Each handling is decomposed primitives of
property accesses or function calls.
For example:
// object-unsued.js
httpGet(url.parse("http://example.org")).then(function (res) {
return res.headers["location"].replace(/^http:/, "");
}).then(console.log);
It can decompose continuous actions of each primitive access as:
// object-decomposed.js
httpGet(url.parse("http://example.org")).then(function (res) {
return res.headers;
}).then(function (handlers) {
return handlers["location"];
}).then(function (location) {
return location.replace(/^http:/, "");
}).then(console.log);
Promises of Q
have support methods of primitive access
as a promise.
By them, the decomposed actions also translate to:
// object.primitive.js
httpGet(url.parse("http://example.org"))
.get("handlers").get("location").post("replace", [/^http:/, ""])
.then(console.log);
promise.get(name)
:
promise.then(function (value) {
return value[name];
})
promise.put(name, value)
:
promise.then(function (value) {
return value[name] = value;
})
promise.del(name)
:
promise.then(function (value) {
return delete value[name];
})
promise.post(name, arglist)
promise.then(function (value) {
return value[name].apply(value, arglist);
})
promise.send(name, arg1, arg2,...)
:
promise.then(function (value) {
return value[name].call(value, arg1, arg2, ...);
})
promise.fapply(arglist)
:
promise.then(function (value) {
return value.apply(undefined, arglist);
})
promise.fcall(arg1, arg2, ...)
:
promise.then(function (value) {
return value.call(undefined, arg1, arg2, ...);
})
promise.fbind(arg1, arg2, ...)
:
(function (promise) {
var args = Array.prototype.slice.call(arguments, 1);
return function () {
var all = args.concat(Array.prototype.slice.call(arguments));
return promise.then(function (value) {
return value.apply(undefined, all);
});
};
})
(note: result of fbind is a function, not a promise)
promise.apply(thisp, arglist)
:
promise.then(function (value) {
return value.apply(thisp, arglist);
})
promise.call(thisp, arg1, arg2, ...)
:
promise.then(function (value) {
return value.call(thisp, arg1, arg2, ...);
})
promise.bind(thisp, arg1, arg2, ...)
:
(function (promise, thisp) {
var args = Array.prototype.slice.call(arguments, 2);
return function () {
var all = args.concat(Array.prototype.slice.call(arguments));
return promise.then(function (value) {
return value.apply(thisp, all);
});
};
})
(note: result of bind is a function, not a promise)
promise.keys()
:
promise.then(function (value) {
return Object.keys(value);
})
The promise.post(name)
or promise.send(name)
convert a method of the promise value to a promise of the method result.
A result of view()
has methods for all methods of the promise value.
You can use view in then
callback of view()
.
e.g.:
// object-view.js
Q.resolve(new Date()).view().then(function (dateView) {
return dateView.toTimeString().then(function (str) {
return /\((.*)\)/.exec(str)[0]
});
}).then(console.log);
The former promise chains are just only single flow style.
Q
supports flow graph for complex flows.
Not only single flow of promise in chain, you can use a promise for multiple flows.
For example, at processing a result text, one flow is to print to log, other flows are to converting.
// chain-concurrent.js
var text = httpGet(url.parse("http://example.biz")).then(function (res) {
return httpGet(url.parse(res.headers["location"]));
}).then(loadBody).fail(function (error) {
return "Not Found";
});
text.then(console.log);
text.post("toUpperCase").then(console.log);
text.post("toLowerCase").then(console.log);
In the code, printing order is not stable beacuse of the end of three flows are independent each other.
Joining multiple flows also supported with Q.all(promiselist)
For example, the former code of print out at the ordered.
// chain-concurrent-join.js
var text = httpGet(url.parse("http://example.biz")).then(function (res) {
return httpGet(url.parse(res.headers["location"]));
}).then(loadBody).fail(function (error) {
return "Not Found";
});
var texts = Q.all([
text,
text.post("toUpperCase"),
text.post("toLowerCase"),
]);
texts.spread(function (normal, upper, lower) {
console.log(normal);
console.log(upper);
console.log(lower);
});
When all of promiselist
are resolved, Q.all(promiselist)
is resolved.
When one of promiselist
is rejected, Q.all(promiselist)
is rejected.
Other case is join multiple accesses:
Q.all([
httpGet("fragment.glsl"),
httpGet("vertex.glsl"),
]).spread(function (fragmentCode, vertexCode) {
return makeShaderProgram(fragmentCode, vertexCode);
});
Q.allSettled(promiselist)
is similar Q.all
.
Q.all(ps)
is a promise of values of ps
,
but Q.allSettled(ps)
is a promise of settled(success or failure)
promise list. The then
callback arg is an array of result object as:
{state: "fulfilled", value: resolvedValue}
- or
{state: "rejected", reason: rejectedError}
Joining with allSettled
is better for execution based programs
that focus on spawning procedures (not use the result values).
(allSettled
is renamed form allResolved
after q-0.9.4).
promise.done()
is for abort the program entirely.
If the promise
is rejected (and not catched error before),
done()
forcely spawn uncaughtable error (e.g.
setTimeout(function () {throw ex;}, 0)
).
On node.js REPL, run Q.reject("uncaught").done()
, then exit with an error.
If error reached to the done()
, you could think it just a programming bug
(not exception state).
fin
is for invoking common process for then
handelrs. e.g.
// promise-fin-flow.js
promise.fin(function () {
console.log("resolved");
}).then(function (value) {
console.log("success " + value);
}, function (error) {
console.log("failure " + error);
});
If promise = Q.resolve("OK")
, prints "resolved\nsuccess OK"
.
If promise = Q.reject("FAIL")
, prints "resolved\nfailure FAIL"
.
fin
callback has no arguments.
promise has two time control methods:
promise.delay(ms)
promise.timeout(ms)
promise.delay(ms)
is a promise that
spawns then
callback after ms
milliseconds. e.g.
// promise-delay.js
Q.resolve("OK").delay(5000).then(console.log)
It prints "OK"
when 5-sec after.
promise.timeout(ms)
is a promise that
spawns reject callback after ms
seconds when not yet resolved.
For example,
// promise-timeout.js
Q.resolve("OK").timeout(5000).then(console.log)
It prints "OK"
just now. But:
// promise-timeout.js
Q.resolve("OK").delay(10000).timeout(5000).then(null, console.log)
It prints [Error: Timed out after 5000ms]
when 5-sec after.
Q
has simple utility for node.js modules.
Q.nfapply(func, args)
Q.nfcall(func, arg1, arg2, ...)
Q.nfbind(func, arg1, arg2,...)
Q.npost(obj, name, args)
Q.nsend(obj, name, arg1, arg2, ...)
It is just for node.js standard IO module callback style, e.g.:
fs.readFile(filename, encoding, function (error, result) {
if (!error) console.log(result);
});
- the callback arg of actions is at the last argument
- the first argument of callback functions is for error
- the second argument is a result of action
(You can not use them for http module because its style is different)
The transformed examples are:
Q.nfapply(fs.readFile, [filename, encoding]).then(console.log);
Q.nfcall(fs.readFile, filename, encoding).then(console.log);
Q.nfbind(fs.readFile)(filename, encoding).then(console.log);
Q.npost(fs, "readFile", [filename, encoding]).then(console.log);
Q.nsend(fs, "readFile", filename, encoding).then(console.log);
ECMAScript5 spritted Identifier and IdentifierName. The IdentifierName allows to use keyword names.
Property names of ECMAScript5 uses IdentifierName (not Identifier). e.g.
var obj = {
new: function () {return Object.create(null)},
};
obj.new();
Q
promise supports new property names that
includes some of ES keywords names.
Q.begin(obj)
is same asQ.resolve(obj)
promise.delete(name)
is same aspromise.del(name)
promise.try(arg1, arg2, ...)
is same aspromise.fcall(arg1, arg2, ...)
promise.catch(callback)
is same aspromise.fail(callback)
promise.finally(callback)
is same aspromise.fin(callback)
There are many promise implementations in JavaScript.
All of them support CommonJS Promises/A interface.
"CommonJS Promise/A" sets standard interface for promise methods:
promise.then(fulfilledHandler, errorHandler, progressHandler)
promise.get(propertyName)
promise.call(functionName, arg1, arg2, ...)
Q
also supports them (progressHandler is just ignored).
Q.when(value)
accepts "CommosJS Promise/A" promise
(which has the then
method).
Q
promise can chain from other implementation of promises.
from jQuery-Deferred
:
// interop-jquery-deferred.js
var Q = require("q");
var $ = require("jquery-deferred");
var otherPromise = $.Deferred();
Q.when(otherPromise).then(console.log);
otherPromise.resolve("OK from other");
from node-promise
:
// interop-node-promise.js
var Q = require("q");
var promise = require("node-promise");
var otherPromise = new promise.Promise();
Q.when(otherPromise).then(console.log);
otherPromise.resolve("OK from other");
from when
:
var Q = require("q");
var when = require("when");
var otherDeferred = when.defer();
var otherPromise = otherDeferred.promise;
Q.when(otherPromise).then(console.log);
otherDeferred.resolver.resolve("OK from other");