Re: Latest commit breaks master

[David Kastrup]

James <address@hidden> writes:

> Well merge does a full doc build and your checkin was in master
> already so must have made it through the make doc in that case.

No, it wasn't.  It was on top of the stuff of Jean-Charles that had
already migrated from staging to master, and this stuff (multiple
commits) on top of master was failing make doc because it got more picky
about abusing postevents as bona fide music.

> That was what was so odd.

I can't totally vouch that Jean-Charles pushed to staging: it could also
have been that I rebased my patches on master and got his material this
way into staging.

But at the time I was fixing the problem, master was apparently not
broken.  And as far as I can tell, Jean-Charles' patch just adds files
to Documentation/snippets/new, and those are not even looked at as far
as I remember until one does an LSR import.

> Anyway, i can enable a full make doc on every test patchy - it isn't
> that huge a deal.

For me, it is about 1h20 as opposed to 15min.  It would be useful for
documentation patches, but it is not like the occasional clogging of
staging is really much of a problem.  The main advantage is that we
avoid situations where people get tense.

> I've set on for now and I'll see if it affects the amount of testing I
> can get done compared to what I do now.

Thanks.  By the way, is the 2-hourly run on right now?  After fixing the
docs and getting a successful make doc run (big time hole), I pushed to
staging again somewhat more than 2 hours ago, so I am somewhat worried
that something bad got in for whatever reason again in spite of full
testing coverage.

-- 
David Kastrup