Display tests results and code coverage

[FedericoCeratto]

Thanks for developing Helix.
It would be really nice if Helix could receive additional diagnostics from external sources to show failed tests.
Similarly it could set colors in the gutter to show code coverage.

An external tool would be responsible for collecting test outputs and coverage and feeding it to Helix.

(I looked around and could not find the same idea elsewhere)

[steinuil]

In #5014 it was mentioned that the Build Server Protocol might fulfill this use case in the future, but it's been 4 years since that post...

Anyway, I've been thinking about something similar for a while, so today I put together a language server that displays pytest failures in-editor using the built-in JUnit XML report that pytest produces:

https://github.com/steinuil/pytest-lsp

I don't think LSP can interact with the gutters to show code coverage (maybe that'll be something that could be done with the plugin system?), but only displaying test failures is not that complicated in principle!

[mrijken]

I have made a keymap to run unittests (pytest in my case) and show the results in new buffer:

[keys.normal]
C-t = [
    ":new",
    ":pipe pytest 2>&1 || true",
]

In the buffer with the results, you can use gf to directly go to the file and line of the failing test.

As a test can take some time, helix is not usable during that time. As a workaround, I am using now:

[keys.normal]
C-t = [
    ":sh pytest 2>&1 > .pytest.out &",
    ":open .pytest.out"
]

So the tests run in the background and I can open the .pytest.out file to inspect the results and us gf to go to the line with failing tests (and rl to reload the tests results when the tests are not done yet).

But BSP will be much better of course.