Re: Patchy and dev/staging

[Graham Percival]

On Fri, Nov 11, 2011 at 04:35:29PM +0100, David Kastrup wrote:
> Peekay Ex <address@hidden> writes:
> 
> > With regard to the conversations we've had about potential issues me
> > pushing HEAD (so to speak).
> >
> > What do you (both) want me to do with patchy? I don't mind running the
> > script and just sending an email to dev saying all is ok (with the
> > last commit that patchy thinks is there).
> 
> The problem is that your version of patchy is a blackbox to me.  I have
> no idea what it does.

Why is it a black box?  as long as James updates his
lilypond-extra repo, patchy will run the commands in
compile_lilypond_test.py

NB: that said, I don't mean to imply that I expect you to work on
that python code.  You're doing a fantastic job of everything
else; there's no reason why you should be working on this as well.

It's really disappoiting that nobody else is interested in
automatic this task.  :(   finding somebody with moderate git
competency and basic python fluency should not be hard.  :(

> As long as the version it announces fit for pushing is the same commit
> as the top of dev/staging, you can push it to master (or try.  If it
> fails, dev/staging needs to be manually rebased, and I don't mind doing
> that).

> If the version of patchy you currently use is stupid enough, that will
> always be the case.  I am just afraid that based on some
> not-so-well-thought-out advice of mine, your version of patchy might try
> to be cleverer.

If James uses patch from lilypond-extra
  e52f265dd585050fe74250fd291d85886b7f653f
then AFAIK patchy is "stupid enough".  It's doing
  git clone --mirror -s %s .git
  git --git-dir=.git config core.bare false
  git checkout origin/master
  git merge --ff-only origin/dev/staging

and then testing the result.  If it fails at any step (be it a git
command, autogen.sh, configure, make, make test, or make doc) then
it bails.

- Graham