Re: [PATCH] bitbake-layers: layerindex-fetch: respect --branch for already-configured layers

[Osama Abdelkader <osama.abdelkader@gmail.com>]

On Mon, Dec 08, 2025 at 09:40:05PM +0000, Paul Barker wrote:
> On Sat, 2025-11-01 at 15:56 +0200, Osama Abdelkader wrote:
> > [YOCTO #7852]
> > 
> > The layerindex-fetch command has two issues when --branch is specified:
> > 
> > 1. If a layer is already in bblayers.conf, the command exits early with
> >    "You already have the requested layer(s)" without checking if the
> >    layer is on the requested branch.
> > 
> > 2. For layers in bblayers.conf (cooker layers), the branch checking code
> >    added in commit 138dd7883ee (2022) never executes because cooker layers
> >    are skipped entirely in the fetch loop.
> > 
> > This patch fixes both issues by:
> > 
> > - Preventing the early exit (fast path) when --branch is specified, forcing
> >   branch verification even for already-configured layers.
> > 
> > - Adding branch check/switch logic for cooker layers when --branch is
> >   specified. The code finds each layer's git repository and calls
> >   get_fetch_layer() with the requested branch to perform the validation
> >   and checkout if needed.
> > 
> > - Fixing the branch detection in get_fetch_layer() to use
> >   'git rev-parse --abbrev-ref HEAD' instead of parsing 'git branch' output.
> > 
> > - Fixing undefined 'stderr' variable (should be 'completed_proc.stderr').
> > 
> > This completes the fix started in commit 138dd7883ee, allowing users to
> > use --branch to switch already-configured layers to different branches.
> > 
> > Signed-off-by: Osama Abdelkader <osama.abdelkader@gmail.com>
> 
> Hi,
> 
> Thanks for the contribution! Apologies for the delay in reviewing it, we've
> been focused on finishing the 5.3 release.
> 
> This looks like a useful change but it's difficult to review in its current
> form. Could you split this into multiple commits? Based on the above detail
> there are 4 different logical changes here so we would prefer each of those as
> a separate patch.
> 
> I'd especially like the completed_proc.stderr fix to be isolated as it
> may need to be backported to a stable release. So this fix should come
> first in the series - best practice is to put patches that fix bugs
> first, then patches that add features.
> 
> I think you'll also need to apply these changes to the bitbake
> repository rather than the poky repository to remove the unnecessary
> 'bitbake/' prefix on the file paths.
> 
> Best regards,
> 
> -- 
> Paul Barker

Thanks for the feedback, I'm going to that.
Best regards,
Osama