Stale github issues

[kkmuffme]

Since issues are piling up and the more issues there are the less likely people are to contribute from my experience, I think the github bot should be updated.
This will also ensure that issues that might have been fixed in the meantime to be automatically closed.

If more than 6 months since the last contribution, have the bot comment

This issue is stale and will be closed within a month. Do you still experience this issue? Please close the issue if it has been resolved. Otherwise, please consider contributing with a PR. Feel free to ask questions here to get you started.

If more than 7 months, auto-close the issue as "Not planned", since it will never get fixed/implemented anyway - and the issue can be reopened, if there is interest in it anyway.

Only exceptions I'd say:

• "easy problems" and "good first issue" -> messages same, but we never actually close those, since these might get picked up by someone and fixed easily, getting people started with contributing.

[orklah]

I'm really not sure about this. I have checked all the issues, from beginning to end at least twice since last year. I made sure they were all relevant and I closed a lot of them when not reproduced or else.

Meaning about all current issues currently open are legitimate. I have a pretty good memory so when I see a new issue that popped up before, I try to find it and close the new one as duplicate. If we closed old ones just to reopen them later, we would end up loosing history (and me trying to describe what has to be done to fix the issue on some discussions)

I agree the situation is different on enhancements. Most of those will probably never see the light of day - unless more contributor steps in. But it's not the fact that they're old, it's just we lack contributors motivated to work on those. Some real good ideas are described in currently opened issues. If we just close all enhancement after a year of inactivity, we'll just close about all of them, this is pretty sad. I prefer having a well maintained backlog than just letting people create issues just for them to be closed by a bot later.

What I would really love to help maintain issues is something phpstan has and I'm quite jealous of it 😄 Their bot will automatically post on each issue whenever the output of the reproducer changed. This basically creates a second CI based on all the currently open issues, each change on the code will get confronted to a lot of cases. It will help find fixed issues automatically and also detect regressions by having a lot of user code that is often on the edge of what's working and what is not.
If I could, I would go even further by forcing users to post code snippets that can automatically detect when fixed. We often have @psalm-trace in comments and a description of what the result should be. If users could actually add an annotation that would pass when the bug is fixed, we could even subcontract our unit tests to users!

[AndrolGenhald]

This is the sort of thing I was aiming to move toward (very incrementally) with #8172. That's neat that PHPStan has this all set up, maybe I'll have to take a look at how they do it.

[kkmuffme]

Oh ok, that's good. I thought those issues are left and forgotten :D

Their bot will automatically post on each issue whenever the output of the reproducer changed

Maybe we can see how they do it and implement it in a similar way?

If I could, I would go even further by forcing users to post code snippets that can automatically detect when fixed. We often have @psalm-trace in comments and a description of what the result should be. If users could actually add an annotation that would pass when the bug is fixed, we could even subcontract our unit tests to users!

That's kind of what rector does: https://getrector.org/demo/61323bb5-4d27-4990-81a0-610b880a5109 click on "Create a Test"
This would be quite simple to implement (we wouldn't even need a PR like they do, just maybe a @psalm-expect comment or similar to be used in psalm.dev so we know what type it should be there)