prompt user to choose parser to parse task output

[elaihau]

• before a task runs, prompt the user to choose which parser to use to parse the task output, and write user's choice into tasks.json. Please note: tasks that are already assign problem matchers
• part of Align UX for tasks with VS Code #4212

How to test

1. Have a detected task prepared. A configured task with no defined problem matcher also works. If a detected task is involved in the testing, make sure it is successfully loaded as Theia currently has a bug where the detected tasks are not loaded in time some plugins are not loaded in time #5861 (comment)

2. Run the prepared task, as a user I should see a list of options that advise me what to use to parse the task output.

• If the task is already assigned a parser, as a user I should not see the list show up.

3. After the task finishes running,

• the problems found by the parser should be listed in the problems view
• the selected problem matcher should be written into the tasks.json as one property of the task.

Review checklist

• as an author, I have thoroughly tested my changes and carefully followed the review guidelines

[RomanNikitenko]

@elaihau
Hello!
I tested your changes using npm extension and faced with the following case:

1. Run npm:build task -> drop down menu -> select eslint item
2. The configuration is moved to tasks.json file and problemMatcher is defined
3. The parser is assigned to the task, so as a user I should not see the list show up at next running
   It works well.
4. What user need to do to get again drop down menu and select another item?

I deleted problemMatcher field for the task in tasks.json file and run again the same task - > got drop down menu -> selected another item

As result, I got two configurations with the same label in tasks.json file and in Terminal -> Run Task menu.
Please see the video: https://youtu.be/zpoZ0F0lXok

Can the behavior be improved?
Like here for example: https://youtu.be/SgBuu1-1X5g
WDYT?

thank you!

[elaihau]

@RomanNikitenko thank you for testing, really appreciated !

Yes it is definitely a bug. What I did to "get the drop down menu and select another item" is removing the entire cutomization object from tasks.json, not just the problemMatcher property. So, i never tried what you caught :)

Fixing it is more difficult than I thought, as we don't have the info of where a taskDefinition comes from.

Worse case i might need to change some interfaces. I will see what is the lowest-impact solution to fix it.

[RomanNikitenko]

What I did to "get the drop down menu and select another item" is removing the entire cutomization object from tasks.json, not just the problemMatcher property.

I noticed that "problemMatcher": [] was added to config when I selected Never scan the task output, so I thought that I need to remove this field to get the drop down menu again :-)
But you are right - maybe the more obvious case is removing the entire customization object...

[elaihau]

For the information of reviews of this pull reuqest: #5915 is created to fix the issue where task definitions are not loaded before the initialization of tasks. It is a blocker of testing and merging this pull request.

[elaihau]

@vince-fugnitto @RomanNikitenko @azatsarynnyy
i did a rebase. this pull request is review-ready

[vince-fugnitto]

It looks really good, nice work @elaihau !
I tested the following:

Observation	Status
ProblemMatchers are sorted alphabetically	👍
When prompted, choosing continue... will not write to the tasks.json	👍
When prompted, choosing never... will write [] to skip parsing	👍
When prompted, choosing a parser will write the parser successfully	👍
Writing the parser works correct in any location in the tasks.json	👍
Writing the parser works correctly event with a trailing comma	👍
Manually removing a ProblemMatcher successfully resets and the user must be prompted again if he re-runs the task	👍
If an appropriate ProblemMatcher is chosen for a given task, the markers are reflected in the problems-view	👍
If a ProblemMatcher exists for a given task, no prompt is displayed and the ProblemMatcher is used successfully	👍
Run Last Task still works correctly, the task is re-run with its defined ProblemMatcher if available	👍

The code also looks fine to me, I don't have any objections :)

[elaihau]

Thank you Vincent ! Your review covered so many things I didn't try out.

[vince-fugnitto]

I opened a follow-up issue to enhance the functionality by supporting adding the ProblemMatchers to the task schema so users can benefit from auto-completion and assistance #6134