Ask user to select PR templates when forking a repository

[babakks]

This PR fixes #143083

[babakks]

Besides this PR's main focus, there's one other thing that I don't quite get it.

Here's the typical scenario that happens: when the Publish to Github command is triggered, which is the only command registered by the extension, it always creates a new repository (here) for the user and then pushes files to it. If the push fails, thenGithubPushErrorHandler comes in and, after some primary checking, tries to create a fork from the repository.

Now, my question is, why do we try to fork the repository? We've already created the repository itself. To me, this seems like a general way of handling push errors, so I checked for any other usage for the GithubPushErrorHandler but couldn't find any. Am I missing something?

To verify my point just check the publishRepostory() method (here).

@joaomoreno @alexr00

[joaomoreno]

Now, my question is, why do we try to fork the repository? We've already created the repository itself. To me, this seems like a general way of handling push errors, so I checked for any other usage for the GithubPushErrorHandler but couldn't find any. Am I missing something?

Yes, if you clone a repository for which you have no permissions to push, this error handler comes in handy and prompts you to fork the repository. Catching a push error is more about that scenario than the Publish to GitHub scenario.

[joaomoreno]

It's very cool that you've added tests... even though they feel a bit overkill.

I've merged main and cleaned the implementation a bit, so we don't create functions all the time. Doing that, I found an issue. On case insensitive systems (ie Windows), a pull_request_template.md file will appear twice: once as pull_request_template.md, and once as PULL_REQUEST_TEMPLATE.md. It would be best to rely on fs.readDirectory on all detection; that way we'd always get what the OS has, instead of assuming file name casing.

[babakks]

Yes. I had missed the case sensitivity point too. I've updated the code and replaced fs.stat with fs.readDirectory.

[babakks]

Speaking of case sensitivity, the new testWorkspace directory contains both pull_request_template.md and PULL_REQUEST_TEMPLATE.md. This can be troublesome for people who clone the repository on Windows. So, I think it's better to keep one of the files/folders in the testWorkspace. Do you agree with me?

[joaomoreno]

Yes, best to remove one of them.

[babakks]

Done. I've deleted lowercase counterparts.

[joaomoreno]

Thanks! 🍻