Contributions on GitHub getting stuck

[Alan Bateman]

On 22/10/2021 00:19, Magnus Ihse Bursie wrote:
> I just noticed that there are many PRs on openjdk/jdk that has gotten 
> stuck in a Catch 22, due to how our rules and Skara tools work together.
>
> This list, which currently has 24 PRs on it, lists open PRs that 
> neither have a "oca" nor a "rfr" label:
>
> https://github.com/pulls?q=is%3Aopen+is%3Apr+draft%3Afalse+repo%3Aopenjdk%2Fjdk+-label%3Arfr+-label%3Aoca+ 
>
>
> It seems that this happens when someone creates a PR, gets told to 
> sign the OCA, does so, but then still not is a OpenJDK author (of 
> course!) and so cannot create a JBS issue. And without a JBS issue 
> number, Skara does not move the PR to the RFR state. And without that, 
> no mail is sent out. And at least for the PRs in the build category, I 
> was completely oblivious to their existence.
>
> Even though we write stuff like "Please talk on the relevant mailing 
> lists first" in the User Guide, people assume (rightfully so, I 
> believe) that submitting PRs is an acceptable way to contribute. And 
> when told by the bots to go sign the OCA and then being told "Nice, 
> you're good to go now!", and still nothing happens, this creates bad 
> optics.
>
> At the very least I think Skara should add some message in these cases 
> saying explicitly "You need a JBS issue too get this PR reviewed. 
> Please ask on the relevant mailing list for sponsorship". And maybe we 
> should rethink if we perhaps should not send a mail in any case, at 
> least notifying folks that there is an incoming PR that might need 
> guidance.
I don't disagree that it would be helpful if a bot added a comment to 
say that a JBS issue needs to be created.

BTW: Of the 24, it looks like 12 (the PRs with the "[PATCH]" prefix in 
the description) are from a prolific contributor that is doing code 
cleanup right across the JDK. He has a JBS login so it might be that 
these are really draft changes.

-Alan