helpers/factory: default formatter valid exit code to 0

[barrett-ruth]

Resolves #216.

Warns on formatter failures, defaulting to != 0.

Create a utility wrapper helpers.check_exit_code to maintain backwards compatibility.

TODO

1. logger:warn spawns a blocking notification (i.e. must press enter) and creates a timeout error if not pressed within the defualt timeout

^ this is unavoidable - addressed by using a shorter initial message pointing you to to the log file.

[barrett-ruth]

@mochaaP I can't assign reviewers so please take a look. I know I need to squash commits and a bunch of other stuff but I just wanted to draft the initial API. Let me know everything that can be improved.

[barrett-ruth]

@mochaaP I'm going to start going through the formatters one-by-one (pain) to ensure I'm passing the right exit code.

1. Is there a better way to do this?
2. I currently make it optional (and default to exit code 0) to indicate formatter success in check_exit_code. Is this fine?

[mochaaP]

I think (2) is fine.

For (1), we could land the changes first and wait for feedback from ones who actually use them if any problem arises.

[barrett-ruth]

Ready for review.

[barrett-ruth]

@mochaaP addressing both comments above:

The problem I have with moving the code to the generic factory (and this includes the code duplication comment above) is that I thought this feature only makes sense for formatters.

If we added this to generator factory, wouldn't we also be checking linting exit codes as well?

[barrett-ruth]

@mochaaP this is ready for review, see above comments

[barrett-ruth]

What does the eyes emoji mean 😭 I've already addressed the comments I thought... what do you think? (lol)

[mochaaP]

“I will take a look at it” 😂 This needs some slight changes, I will take care of it.

[mochaaP]

Do you think this implementation is okay?

[barrett-ruth]

@mochaaP you updated the logging message to use logger:add_entry.

Then you deleted the file containing that change (check_exit_code) - lol! I'm wondering if this was intentional? I thought we wanted this informative message to the end user?

If this is the case, I thought we had to separate the logic out in another file to do that log... because only in the formatters does the notion of validating the exit code make sense.

Sorry this is taking so long, LMK if you know what I'm saying.

[mochaaP]

ack, will check tonight.

[barrett-ruth]

LGTM!

[geril2207]

@mochaaP @barrett-ruth
prettierd nvim(nightly but idk if it changes anything):

2025-01-09.12-19-02.mp4

before:

2025-01-09.12-19-43.mp4

[mochaaP]

cc @barrett-ruth is prettierd outputting to stdout? we might need to ignore that

[daephx]

Found myself here because I updated none-ls and keep getting the error nvim_echo must not be called in a lua loop callback when the buffer has invalid syntax. I've only tested with stylua, and haven't checked other formatters.

The error output itself is rather annoying, so for now my solution is to pin none-ls to commit 83c7c00.

Edit: Ok, I just spotted #237 which has the same changes I concluded on after exploring the traceback.

[mochaaP]

I've partially reverted this until I figure out a better way to present the error message.