[Meta] Issue tracker quality

[strugee]

Problem: the issue tracker is often full of low-quality submissions that need work to clean up. (Sometimes when I have some extra time I go through and clean obvious duplicates, bad formatting, etc. - I assume this is helpful but if not please let me know!)

Proposed solution: I think there are a couple things we could do to improve overall issue tracker quality.

1. People seem to like submitting empty issue reports with titles like "bug". It seems like these virtually never turn into actionable issues, so they're mostly a waste of time. I propose we add a note in the issue templates that says that if the template isn't filled out, issues will be automatically closed. We can then enforce this with https://github.com/rohitjmathew/probot-template-enforcer or maybe https://github.com/marketplace/actions/template-validator (I haven't really looked into the quality of either of these apps since we haven't decided to use them yet).

2. I also propose we shorten the "stale" time to 2 weeks instead of 4, provided that we figure out a way to automatically remove the "needs info" label when someone comments. This is because I looked into recently closed stale issues and there were basically two categories: issues where the original submitter never came back and provided the necessary information (non-actionable and useless), and issues where there was actually a productive discussion, but it just kind of fell off everybody's radar and the bot closed it.

Shortening the "stale" time to 2 weeks would still give 4 weeks total for the original poster to comment which seems like plenty, especially given how infrequently these issues actually become actionable. Automatically removing the "needs info" label would solve the second category above and prevent those from being swept up even more in the shorter 4 week stale bot time frame.

/cc @tobiasKaminsky, @AndyScherzinger and feel free to loop in other people as well; I don't know who spends a lot of time in the Android app these days

[strugee]

Actually, a third thing: if we could find a way to detect reports of identical crash signatures, that would help a lot. From what I can tell crash reports are the #1 cause of duplicate issues.

It also might be helpful if we could write a bot to detect common bad issue titles, like "bug" or "Android crash" and leave a comment asking the poster to retitle?

[AlvaroBrey]

Template validation and improving stale behaviour sound great to me. They should not be a great deal of work to implement and would definitely make it easier to maintain the repo.

Detecting crash signatures would be amazing, but I think it's also pretty hard to implement and would need some fuzzy logic for stacktrace analysis. For example, in #8862 you can find a lot of duplicates of the same crash (sidebar image too big) but each one is different because the image size is different.

[tobiasKaminsky]

Template validation and improving stale behaviour sound great to me. They should not be a great deal of work to implement and would definitely make it easier to maintain the repo.

Good idea from @strugee, let us start "simple" by doing this 👍

[AlvaroBrey]

RE: bot behaviour: We recently merged #9190, which should allow us to disable stale bot for an issue even if it has the needs info label. This allows us to mark issues that don't have a clear resolution but shouldn't be closed because they affect users.

[joshtrichards]

People seem to like submitting empty issue reports with titles like "bug". It seems like these virtually never turn into actionable issues, so they're mostly a waste of time.

Empty reports have fallen off dramatically based on a quick analysis of the past 12 months. Likely in part due to #9782 / #10528

provided that we figure out a way to automatically remove the "needs info" label when someone comments

We need to add labels-to-remove-when-unstale: 'needs info' to our stale.yml.

https://github.com/actions/stale#labels-to-remove-when-unstale

Since this workflow is shared (via the android-config repo) I'll submit the PR over there.