DMD PR management hits a new low

Michael V. Franklin slavo5150 at yahoo.com
Sat Nov 18 07:52:43 UTC 2017


I'll just refer you to this comment: 
https://github.com/dlang/dmd/pull/6947#issuecomment-345423103

> Manually merging this pull as it sat around long enough waiting 
> to be marked approved that it accumulated github's max 1000 
> status updates per commit id and won't ever see more until a 
> new commit becomes current for it. Let that sink in... over 
> 1000 builds done for a single pull request before it got marked 
> for merging.

When this happens the auto-tester can no longer update the status 
of the PR.  There are other PRs also suffering from this problem, 
just to name two:

https://github.com/dlang/dmd/pull/6435
https://github.com/dlang/dmd/pull/6260

Since the auto-tester can't update its status, it appears as 
though the auto-tester has stalled with the last status update:
e.g
#6435: auto-tester — Pass: 3, In Progress: 2, Pending: 5
#6220: auto-tester — Pending: 10

I'm not sure what to do about this at this time.  The PRs could 
be woken up by the original contributor, but submitting a new 
commit, but it appears some of the contributors have lost 
patience and moved on.  I'm trying to revive a few, but will my 
PRs suffer in similar fate?

I'm sorry for being negative, but this is shameful, IMO.  Some of 
the PRs are very simple, and just need someone with authority to 
make a decision (e.g. #6435 and #6260)

I'm thinking about creating a monthly list of neglected PRs and 
submit them to the forum on the 1st of every month to bring some 
attention to them, but it seems that same information is already 
available through Github's interface, and I don't know if such an 
effort would be effective.  But, when the auto-tester can no 
longer update the PR's status, it make reviewing and making a 
decision about them, *much* more of a hassle.  We need to prevent 
PRs from getting to that state.

Can we get some resources allocated to this, please?  What can I 
do?

Thanks,
Mike



More information about the Digitalmars-d mailing list