zebiano / labeler Goto Github PK

This project is completely blind in terms of testing. Whoever works on it and does some changes has no idea if it breaks something not possible to be tested. Ideally, tests would be created and ran before releasing a new version, and/or when opening PRs, etc.

I was checking through my code and realized it could be a bit more organized. Sort of like a "rework" of the way code is written.

I've encountered a bug where labels.json gets saved to the current directory, instead of the package directory. I'm not sure if it could even be considered a bug, as some users might actually like it that way. Though it was not intentional and I personally don't like it.

Version @2.1.1 uses current directory, version @2.1.4 uses package directory. The drawback of using the package directory is having to run labeler with sudo, since fs needs root permissions to write into the package directory.

I'm still unsure on how I'll handle this situation. Pick the version that suits you best for now.

This is already implemented for GitHub Enterprise (flag -b, --bulkUpdate), but it would be nice to also have it implemented for a normal GitHub user.

Wanted is for the -b flag to also work with non-GHE instances. It should update every repository that is fetchable through the API of an User and apply the desired Labels.

This is something that crossed my mind as the token is quite confidential. If enough users want this, I could take it into consideration.

This way, whenever you enter the token through the labeler -c config, it'll be hidden.

Very simple, and something I will most likely develop.

When running labeler -n, there might be some users who expect to start a fresh new file (which is not the case currently). So by asking if the user wants to start fresh, that is that problem solved.

I might also make the -f work with it, and I guess it'll default to "No, I don't want to reset the file". We'll see.

Whenever an error happens, it's displayed like:

Error: An unexpected error occurred.
Error: Error: Request failed with status code 401 

Ideally, the word Error should not get repeated twice.

With labeler -n, only D2DAE1 is seen as a valid input, and not #D2DAE1. When trying to use #D2DAE1, the error doesn't tell the user to remove the hashtag, but instead simply says "Please enter a valid Hex color".

I'm starting to get scared of breaking existing features, so having automated tests for each function would be nice.

I think this is a fantastic tool, but I can't figure out how to get it working with a GitHub Enterprise instance. Is that currently supported, or do you know an easy way to help make that happen?

GitHub recommends using the following header when making a request:

accept: application/vnd.github.v3+json

Basically specifies that the answer should come in json format (which I guess is the default one anyways), as well as the API version. This ensures labeler will always work with the GitHub API, at least with the v3 version.

This is actually quite important. Whenever no labels.json file is detected, a new one gets created with the default labels that are stored in the "lib/fs.js" file. Though currently, if a user updates labeler and the default labels were changed, labels.json will not get updated.

This is actually planned, since I don't want to accidentally overwrite a custom labels.json file from users. But I also want to give the option to manually update the file to the defaults. As such, I recommend a --dlf flag (delete labels file).

Hello. Thanks for the Tool.

It seems that CTRL+V is broken. I can not paste Text I have copied to the Clipboard to

• Enter Label Name
• Enter Label Description
• Enter Label Color

Is this a limitation of the Script or of NPM?

• Windows 10 21H1
• NPM 7.18.1
• Node 16.4.1

Thanks!