Support JupyterLab 2.2 #301

krassowski · 2020-07-29T01:12:59Z

References

Supersedes #239, fixes #241.

Code changes

adopted to ICompletionItem interface
implements ICompletionItemsConnector over simple DataConnector
uses kernel icon for suggestions from kernel if type is not available
passes documentation and deprecation note to the completer

User-facing changes

In addition to the new shiny completer widget (see all three PRs for jupyterlab/jupyterlab#7983):

the autocomplete is no longer overaggressive (Only Autoinsert Single Completion Items on Tab jupyterlab/jupyterlab#8375) and does not cause path-completion issues - this opens way to implementing hitlander mode
- this means that pressing dot . once does not cause autocompletion to insert text if only one item was found. Pressing tab twice does the trick.
we get label rather than insertText displayed

Backwards-incompatible changes

ICompletionItemsConnector is a breaking change and not backwards compatible with JupyterLab 2.1

Chores

linted
tested
documented
changelog entry

…-lsp into kiteco-completion-items

krassowski · 2020-07-29T01:15:21Z

Note: disable pre-commit hook for now as it was destroying code and could not work around it.

@bollwyvl I am prettier illiterate; it is breaking completions.ts (literally does mish-mash) when run and I have no idea how to fix it, nor which piece it does not like. How do I lint/fix this?

bollwyvl · 2020-07-29T01:48:21Z

You can disable it explicitly for a single file in .prettierignore. Might be sufficient for this case. There are also some pragmas we can enable, which would let you put it magic comments. But it'd be best if we could fix through normal means, no doubt.

Anyhow, can you post an example of before/after what it's doing as a gist? Maybe we need to upgrade? Please, let's not downgrade, as that would ruin my world on #278, which I'm already scared of fixing the bitrot on.

But indeed, who lints the linters! I disable the hooks, too, because I don't like node doing anything i don't type out with node, yarn or jlpm... git is for me.

packages/jupyterlab-lsp/src/adapters/jupyterlab/components/completion.ts

bollwyvl · 2020-07-29T02:28:31Z

yarn.lock

-  version "7.8.3"
-  resolved "https://registry.yarnpkg.com/@babel/code-frame/-/code-frame-7.8.3.tgz#33e25903d7481181534e12ec0a25f16b6fcf419e"
-  integrity sha512-a9gxpmdXtZEInkCSHUJDLHZVBgb1QS0jhss4cPP93EW7s+uC5bikET2twEF3KV+7rDblJcmNvTR7VJejqd2C2g==
+"@babel/code-frame@^7.0.0", "@babel/code-frame@^7.10.4":


it may well be time to do a full re-solve of all our deps. We should probably do some better sanity pins on, for example, the node-based language servers as well.

surprisingly, seems to be largely working :) the most important trick was in krassowski@bf52857

krassowski · 2020-07-29T07:35:30Z

Also, why didn't Azure CI fire?

krassowski · 2020-07-29T07:39:30Z

Well, seems that running jlpm prettier from command-line worked better than from a hook...

krassowski · 2020-07-29T15:32:07Z

All completer tests now pass locally! CI (https://dev.azure.com/krassowskimichal/jupyterlab-lsp/_build) has issues with backend lint:

source activate jupyterlab-lsp && python scripts/lint.py
========================== Starting Command Output ===========================
/bin/bash --noprofile --norc /home/vsts/work/_temp/3d27c3e5-264b-4ba5-8197-7ce2012605f5.sh
/usr/share/miniconda/envs/jupyterlab-lsp/lib/python3.7/site-packages/setuptools/distutils_patch.py:26: UserWarning: Distutils was imported before Setuptools. This usage is discouraged and may exhibit undesirable behaviors or errors. Please use Setuptools' objects directly or at least import Setuptools first.
  "Distutils was imported before Setuptools. This usage is discouraged "
All done! ✨ 🍰 ✨
52 files left unchanged.
py_src/jupyter_lsp/tests/test_paths.py:35: error: Argument 1 to "__call__" of "_SkipifMarkDecorator" has incompatible type "int"; expected "Union[str, bool]"
py_src/jupyter_lsp/tests/test_paths.py:61: error: Argument 1 to "__call__" of "_SkipifMarkDecorator" has incompatible type "int"; expected "Union[str, bool]"
Found 2 errors in 1 file (checked 39 source files)

which seems like a good mypy catch because ~WIN would mean:

>>> ~True
-2
>>> ~False
-1

so we should go for not WIN. Is this correct @bollwyvl?

bollwyvl · 2020-07-29T23:48:51Z

Yeah, not WIN looks correct.

krassowski · 2020-08-06T14:46:40Z

Hooray, azure kicked in after merging master this time!
Not sure if I have yarn lock right, there were conflicts and I just resolved it anew after merging... May be missing something - let's see how it plays out on CI

bollwyvl · 2020-08-07T13:35:33Z

I've done some digging on that error code before... Sometimes it's enough to ensure the thing being selected by this keyword is visible, _and_ unique... Will dig into it this weekend, might have to rebuild my windows VM.

krassowski · 2020-08-07T13:36:47Z

Now the question is why

MacOSXThreeSeven and
WindowsThreeSeven, WindowsThreeEight

fail while MacOSXThreeSix, WindowsThreeSix, MacOSXThreeEight and all Linux tests pass. I see no pattern...

krassowski · 2020-08-07T13:37:45Z

Well synced, your message appeared 0.5 second before I pressed enter ;)

bollwyvl · 2020-08-07T15:46:16Z

Also, the win 36 is kind of "best effort" because of ignoring some previous pyls issues... So it's not really an indicator for a lot of features. Will revisit...

…

On Fri, Aug 7, 2020, 09:37 Michał Krassowski ***@***.***> wrote: Well synced, your message appeared 0.5 second before I pressed enter ;) — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <https://github.com/krassowski/jupyterlab-lsp/pull/301#issuecomment-670520871>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAALCRHU25QGXR5JWELG763R7P7TPANCNFSM4PLCQU2Q> .

krassowski · 2020-08-10T10:43:43Z

Now archspec (whatever this is) also fails to conda install on Windows...

ERROR conda.core.link:_execute(568): An error occurred while installing package 'conda-forge::archspec-0.1.1-pyh9f0ad1d_0'.
CondaError: Cannot link a source that does not exist. C:\Miniconda\Scripts\conda.exe

bollwyvl · 2020-08-10T20:25:05Z

On windows: cannot reproduce the archspec thing (not familiar with it), but can consistently reproduce the rect error, so probably will be able to fix it... or at least determine why we should allow it ignoring it.

bollwyvl · 2020-08-10T23:57:11Z

atest/05_Features/Completion.robot

@@ -108,6 +130,7 @@ Cell Editor Should Equal
 Select Completer Suggestion
    [Arguments]    ${text}
    ${suggestion} =    Set Variable    css:.jp-Completer-item[data-value="${text}"]
+    Wait Until Element Is Visible   ${suggestion}  timeout=10s


Hopefully this works! 🤞

krassowski · 2020-08-11T00:13:31Z

ci/steps.conda.yml

@@ -1,7 +1,7 @@
 parameters:
  name: Linux
  packages: ''
-  install_cmd: conda install -yn base -c conda-forge conda python-libarchive-c
+  install_cmd: conda install -yn base conda


I guess this is an attempt to tackle the archspec issue?

yes, the thinking is get something as vanilla as possible, especially for windows, while we get this sorted. who knows what state it's in when they cook the docker images? We could also pin harder, e.g. 4.8 or something.

if we want to start getting far off vanilla, we could investigate installing mamba in this step instead of upgrading conda... no doubt it could shave off some minutes, at the price of sometimes giving different solves. definitely a separate PR, though!

i also need to investigate caching (or packaging) the tectonic stuff.

Was just thinking about mamba while reading this

I'll open an issue...

https://github.com/krassowski/jupyterlab-lsp/issues/311

bollwyvl

LGTM

edzkite and others added 4 commits April 3, 2020 17:02

Integrate ICompletionItems

8290f3f

Update ICompletionItems calling pattern

dddad10

Merge branch 'completion-items' of ssh://github.com/kiteco/jupyterlab…

8c9ca6f

…-lsp into kiteco-completion-items

Attempt to support JupyterLab 2.2

00216c4

bollwyvl reviewed Jul 29, 2020

View reviewed changes

packages/jupyterlab-lsp/src/adapters/jupyterlab/components/completion.ts Outdated Show resolved Hide resolved

bollwyvl reviewed Jul 29, 2020

View reviewed changes

krassowski closed this Jul 29, 2020

krassowski reopened this Jul 29, 2020

Prettier completion.ts

ec92128

krassowski added 5 commits July 29, 2020 10:55

Fix the versions integrity

e077218

Fix "Module not found: Error: Can't resolve 'net'"

bf52857

Actually use ICompletionItemsConnector :)

444e337

Try to restore prefixing

811d994

Fix loose ends in completer and its tests

c6a0ac7

krassowski mentioned this pull request Jul 29, 2020

Fix tests/mypy error for "not WIN" condition #305

Merged

Lint robot completion tests

b6fff6c

krassowski requested a review from bollwyvl August 6, 2020 14:34

Merge branch 'master' into kiteco-completion-items

15726b5

krassowski added 2 commits August 6, 2020 16:00

Add changelog entry for JupyterLab 2.2 support

4cad202

Lint changelog

a1940e4

krassowski closed this Aug 7, 2020

krassowski reopened this Aug 7, 2020

krassowski closed this Aug 10, 2020

krassowski reopened this Aug 10, 2020

wait for completions to be visible before hover

d2fe5bd

bollwyvl reviewed Aug 10, 2020

View reviewed changes

use conda from defaults

cb5ad3e

krassowski commented Aug 11, 2020

View reviewed changes

bollwyvl approved these changes Aug 11, 2020

View reviewed changes

krassowski merged commit f305ab4 into master Aug 11, 2020

krassowski mentioned this pull request Aug 11, 2020

Autocompleter with documentation and sorting based on LSP suggestions #21

Closed

krassowski deleted the kiteco-completion-items branch September 4, 2020 16:19

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Support JupyterLab 2.2 #301

Support JupyterLab 2.2 #301

krassowski commented Jul 29, 2020 •

edited

Loading

krassowski commented Jul 29, 2020

bollwyvl commented Jul 29, 2020

bollwyvl Jul 29, 2020

krassowski Jul 29, 2020

krassowski commented Jul 29, 2020

krassowski commented Jul 29, 2020

krassowski commented Jul 29, 2020

bollwyvl commented Jul 29, 2020

krassowski commented Aug 6, 2020

bollwyvl commented Aug 7, 2020 via email

krassowski commented Aug 7, 2020

krassowski commented Aug 7, 2020

bollwyvl commented Aug 7, 2020 via email

krassowski commented Aug 10, 2020 •

edited

Loading

bollwyvl commented Aug 10, 2020

bollwyvl Aug 10, 2020

krassowski Aug 11, 2020

bollwyvl Aug 11, 2020

bollwyvl Aug 11, 2020

krassowski Aug 11, 2020

bollwyvl Aug 11, 2020

bollwyvl Aug 11, 2020

bollwyvl left a comment

Support JupyterLab 2.2 #301

Support JupyterLab 2.2 #301

Conversation

krassowski commented Jul 29, 2020 • edited Loading

References

Code changes

User-facing changes

Backwards-incompatible changes

Chores

krassowski commented Jul 29, 2020

bollwyvl commented Jul 29, 2020

Choose a reason for hiding this comment

Choose a reason for hiding this comment

krassowski commented Jul 29, 2020

krassowski commented Jul 29, 2020

krassowski commented Jul 29, 2020

bollwyvl commented Jul 29, 2020

krassowski commented Aug 6, 2020

bollwyvl commented Aug 7, 2020 via email

krassowski commented Aug 7, 2020

krassowski commented Aug 7, 2020

bollwyvl commented Aug 7, 2020 via email

krassowski commented Aug 10, 2020 • edited Loading

bollwyvl commented Aug 10, 2020

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

bollwyvl left a comment

Choose a reason for hiding this comment

krassowski commented Jul 29, 2020 •

edited

Loading

krassowski commented Aug 10, 2020 •

edited

Loading