Re: [RFC] Sort #includes in gdb

[Sergio Durigan Junior <sergiodj@redhat.com>]

On Wednesday, April 03 2019, Tom Tromey wrote:

> Sergio> I'm still investigating, but I think the "--branch" option doesn't
> Sergio> specify a branch to be tested; rather, it specified a branch where the
> Sergio> test should be performed.  I.e., you still have to provide a patch.  We
> Sergio> currently just support testing against origin/master (that's why, on
> Sergio> .buildbot/options, we specify the "try_branch" variable as being
> Sergio> "origin/master").
>
> Yeah, what I did is push my branch, then commit a trivial change and use
> "buildbot try --branch origin/users/tromey/...".

The problem with that is that BuildBot needs to know which branch to use
when comparing builds.  If you specify a branch that's not
origin/master, then there are no previous results to compare against.

Maybe it could work if you created a new branch, waited 1 hour or so
(because our BuildBot doesn't fetch directly from sourceware.org;
instead, it uses a mirror, so you have to wait until your branch
propagates), commited a trivial change, built it, then put the huge
patch on top of it, then put another trivial patch on top, and built it
again.  Not very cool.

> I think it's fine to just test & land this patch in pieces.  Normally
> patches aren't so big.

Again, sorry about the limitations of BuildBot.  I remember spending a
considerable amount of time trying to increase the patch size that's
accepted, but unfortunately the change would have to be propagated to
every buildslave, which is not feasible.

Thanks,

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/