Re: [PULL 0/4] OVMF patches for 2021-07-12

[Daniel P . Berrangé]

On Tue, Jul 13, 2021 at 12:35:18PM +0200, Philippe Mathieu-Daudé wrote:
> On 7/13/21 12:25 PM, Peter Maydell wrote:
> > On Mon, 12 Jul 2021 at 16:02, Philippe Mathieu-Daudé <philmd@redhat.com> 
> > wrote:
> >>
> >> The following changes since commit 
> >> bd38ae26cea0d1d6a97f930248df149204c210a2:
> >>
> >>   Merge remote-tracking branch 'remotes/rth-gitlab/tags/pull-tcg-20210710' 
> >> into staging (2021-07-12 11:02:39 +0100)
> >>
> >> are available in the Git repository at:
> >>
> >>   https://github.com/philmd/qemu.git tags/fw-edk2-20210712
> >>
> >> for you to fetch changes up to d1e79210457323b225c369fa00e97577e0d0da95:
> >>
> >>   MAINTAINERS: remove Laszlo Ersek's entries (2021-07-12 16:55:23 +0200)
> >>
> >> ----------------------------------------------------------------
> >> Patches related to EDK2/OVMF
> >>
> >> - MAINTAINERS: remove Laszlo Ersek's entries
> >> - Introduce X86_FW_OVMF Kconfig symbol
> >> - pc_system_ovmf_table_find: Assert that flash was parsed, document
> >>
> >> ----------------------------------------------------------------
> > 
> > So, even though this pullreq doesn't seem to be changing gitlab CI
> > config, I get a "yaml invalid" failure in the pipeline:
> > https://gitlab.com/qemu-project/qemu/-/pipelines/335995843
> > 
> > "'build-edk2' job needs 'docker-edk2' job but it was not added to the 
> > pipeline"
> > 
> > Any ideas what that's about?
> 
> For me this is related to what I tried to fix last year but
> couldn't convince Daniel there was a problem. See this thread:
> https://www.mail-archive.com/qemu-devel@nongnu.org/msg773939.html

The situation was different back then, as there was no direct 'needs'
relation between the jobs.

This is also addressing a different issue, which I better understand
now. The filtering based on file names is unreliable when combined
with git pushes. Gitlab looks at the files modified in the delta beween
what is pushed as what already exists on the remote branch. So if you
push 3 changes to a branch, and then push another 2 changes later,
files modified by the first 3 changes won't be considered by the filter
rules, leading to unexpected execution behaviour.

Essentially filtering based on filename is only sane when used with
merge requests, not pushes, because a merge request has a well defined
base commit against which file changes are evaluated.

Regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|