Re: Patchy email

[David Kastrup]

James <address@hidden> writes:

> I don't know the ramifications of running two merges at the same time

None, really.  The worst is when the second merge starts from a later
version of origin/staging and still finishes first.

In that case, the first merge will flag an error when it tries pushing
an already superseded master.

Alternatively bad is when a merge starts working on some version
origin/staging that gets removed/overwritten for a different reason than
failing verification.  In that case, when the merge finishes, it will
push something to master that will not match the changed staging.  As a
consequence, the automatism comes to a stillstand until the situation
has been salvaged by basing staging off the unfortunate master and
fixing the problem in staging.  But that problem case is not related to
multiple instances of merges.

> so if you can keep it an hour either side of 18:00 BST to NOT compile
> we won't hit each other.

We've dumbed down the procedure enough that nothing worse than an error
report should be able to happen automatically.

-- 
David Kastrup