Modernize repo and make sure it works with ST4

[Frizlab]

That being said:

• I have no idea if it works before ST4 now It does not work before ST4;
• I have not succeeded in running the tests (UnitTesting shows a panel at the bottom, but it is desperately empty; I suspect a syntax error, or something of that nature, but I don’t know for sure and am not sure how to debug…) -> Fixed;
• I have removed the support for vim modelines. There is a dedicated plugin for that if needed; let’s not mix everything up! -> I actually enhanced the support for it, as well as added Emacs modelines;
• I migrated the EoL from dos to unix;
• I removed things that I thought were useless; I hope they actually were!

[Frizlab]

Okay. The tests run now (I spent quite a lot of time discovering I had Modelines installed twice, somehow…). I have fixed the obvious syntax errors. Now the tests do not pass, but at least I have a working environment to fix the issues.

[Frizlab]

I also intend on completely changing the algorithm for detecting a modeline.
Currently we’re using the comment char that we request from the view, but I intend on using a static token-based recognition.

The rationale for that is we allow changing the syntax of the file, which means the comment char will probably incorrect if the syntax of the file is incorrect…

I think we could probably have the tokens like so (full-line regex):

.+~~\s*(st|sublime):.*~~[^~]*

(We assume here no comment will need to end with a ~ character.)

That would make any of these modelines recognized:

#~~st: x_syntax swift~~

# ~~st:x_syntax swift~~

// ~~sublime: x_syntax swift~~

↓ Syntax is `swift`, no trailing space. Keys and values are always trimmed. (Same for `value`, obviously.)
// ~~sublime: x_syntax swift ; key value ~~

↓ Preferred syntax.
/*~~ sublime: x_syntax swift ~~*/

And those would not be recognized:

↓ No `~~` at the end.
# ~~st:x_syntax swift ~

↓ Missing `~` before `st`.
# ~st: x_syntax swift ~~

↓ Space after `st`.
# ~~st : x_syntax swift ~~

What do you think?

EDIT: Syntax changes.

[michaelblyons]

You don't think \p{P}{1,4}\s*(st|sublime): is good enough?

[Frizlab]

It’s very edge-casy, but there is dnl.

[michaelblyons]

If you want to check ST3 compat, UnitTesting has a GH Action that can do it for you. There's probably not any divergence between platforms, so ST3, ST4 on Ubuntu is probably enough.

If you have CI minutes to burn. 😅

[Frizlab]

Yes I noticed the UnitTesting repo has a GitHub Actions for testing in the CI and thought about setting that up.

Thanks for the review! I have not finalized everything yet, the PR is still a draft for now (I have to implement the format change, cleanup the tests even more, etc.) I will do the changes you proposed/requested before marking the PR ready.

[Frizlab]

I may be over-engineering things and have 0 time, but I’m still on it! \o/

[michaelblyons]

I'm still loosely watching your progress. (Great work, btw!)

If you hadn't noticed, some time in the last two days, the CI started failing without test output.

[Frizlab]

Thanks!

Yes, I know, I intend to fix that before marking the PR ready.

I'm missing the parser for the VIM and classic modelines, as well as fixing and adding some tests and then I'll be good!

[Frizlab]

The core is done.
Now on to fixing and adding a few tests, and we should be good!

[Frizlab]

Hi @michaelblyons!
I’m apparently hitting a road-block now.

Locally the few tests I have restored work. Using UnitTesting/sbin/run_tests.py, the tests also work!
However, using the GitHub Actions, I get no output at all.

I’m not sure how to debug that. Do you have some insight on how I could proceed?
I’ll continue working on the tests locally in the meantime.

Thanks

[michaelblyons]

Does it help you to debug if you set capture_console to true? Either with the CI yaml or with unittesting.json?

Then you should be able to see any print statements (or logging?) and see how far your program executed.