[Mesa-dev] [PATCH mesa] build systems: uniformize git_sha1.h generation

[Eric Engestrom]

On 28 June 2017 09:20:27 BST, Emil Velikov <emil.l.velikov at gmail.com> wrote:
> On 27 June 2017 at 13:47, Eric Engestrom <eric.engestrom at imgtec.com>
> wrote:
> > Signed-off-by: Eric Engestrom <eric.engestrom at imgtec.com>
> > ---
> > Note: Autotools and SCons are tested, but Android isn't.
> > ---
> >  git_sha1_gen.sh                      | 13 +++++++++++++
> >  src/Makefile.am                      | 13 +------------
> >  src/SConscript                       | 28
> ++++++++--------------------
> >  src/mesa/Android.libmesa_git_sha1.mk |  7 +------
> >  4 files changed, 23 insertions(+), 38 deletions(-)
> >  create mode 100755 git_sha1_gen.sh
> >
> > diff --git a/git_sha1_gen.sh b/git_sha1_gen.sh
> > new file mode 100755
> > index 0000000000..9630067be0
> > --- /dev/null
> > +++ b/git_sha1_gen.sh
> > @@ -0,0 +1,13 @@
> > +#!/usr/bin/env bash
> > +set -eu
> > +
> > +# run git from the sources directory
> > +cd "$(dirname "${BASH_SOURCE[0]}")"
> That's a cool way of avoiding the  --git-dir bits
> 
> This can become cd "$(dirname "$0")" and thus s/bash/sh/ across the
> board?

Sure, changed locally.

> 
> > +
> > +# don't print anything if git fails
> > +if ! git_sha1=$(git rev-parse --short=10 HEAD)
> Error messages seems to be printed - see below examples. A simple
> 2>/dev/null redirect should do it ?

I meant "don't print the git_sha1.h contents"; I think errors are useful here.
If you feel strongly about it, I can silence them.

> 
> git: command not found
> 
> fatal: Not a git repository (or any parent up to mount point /)
> Stopping at filesystem boundary (GIT_DISCOVERY_ACROSS_FILESYSTEM not
> set).
> 
> With the latter message in mind, perhaps we should set --git-dir
> regardless, since git will otherwise traverse all the way to /? Might
> want to leave that for another day, though.

Sure, adding `--git-dir=.git` since we should already be in
the right dir anyway, with the above `cd`.

> 
> With the bash/sh and error messages silenced
> Reviewed-by: Emil Velikov <emil.velikov at collabora.com>
> 
> Thanks
> Emil