Bot reusing branch names?

[erik.joelsson at oracle.com]

Found the issue, filed https://bugs.openjdk.org/browse/SKARA-2242

/Erik

On 4/25/24 05:40, erik.joelsson at oracle.com wrote:
> On 4/25/24 05:34, erik.joelsson at oracle.com wrote:
>> Hello Johan,
>>
>> (I finally had some time to look at this today)
>>
>> I recently changed the fork repository that the bot should be 
>> creating these PRs from. It used to be "openjdk-bot/<repo>" but it's 
>> now "openjdk-bots/<repo>", note the "s". In this PR the actual source 
>> repo is the old "openjdk-bot" but the body text has "openjdk-bots" in 
>> the instructions. The branch name is calculated based on the number 
>> of branches present in the fork, so the number 2 indicates that it 
>> was based on the new "openjdk-bots" repo, but still the PR has 
>> "openjdk-bot/mobile" as the source. I'm puzzled and don't know how 
>> this could have happened.
>>
> I believe this was caused by me changing the bot configuration, but 
> this bot had cached local repositories in its private scratch work 
> area, and those cached repositories had remotes still pointing to the 
> old fork. I have deleted all cached data, and restarted the bot, but 
> this PR is still broken and need to be closed.
>
> /Erik
>
>> /Erik
>>
>> On 4/25/24 00:25, Johan Vos wrote:
>>> Until recently, all PR's from the openjdk-bots that were created to 
>>> allow solving conflicts when merging commits from openjdk/jdk:HEAD 
>>> into openjdk/mobile:HEAD went fine.
>>> The last PR that was created though [1], seems to be weird and I 
>>> wonder if there is something wrong with the PR generation.
>>>
>>> From the github website, it seems there are 25 files with a 
>>> conflict, although that does not seem to be true when I compared 
>>> them manually.
>>> The usual process of fixing the merge conflicts and pushing them to 
>>> the openjdk-bots/mobile:+2 branch did not trigger the next steps in 
>>> the PR on openjdk/mobile.
>>>
>>> What surprises me is that the latest branch in the 
>>> openjdk-bots/mobile repository has the name "2". In the past, those 
>>> branches had incremental names, and the previous branch was named 
>>> "14". I expected to see a branch named "15" but the PR was raised 
>>> from a branch named "2" which is 27772 commits behind the master of 
>>> openjdk/mobile.
>>>
>>> - Johan
>>>
>>> [1] https://github.com/openjdk/mobile/pull/22
>>> [2] https://github.com/openjdk-bot/mobile/tree/14