Proposal regarding handling of kernel configs

[Elliott Mitchell]

On Wed, Apr 26, 2023 at 01:14:54PM -0700, Elliott Mitchell wrote:
> I propose changing how kernel configuration files are handled in Git.
> Specifically, I propose all targets on the development branch should
> *always* have "config-<oldver>" and "config-<newver>" files.  When
> development moves off the older kernel, the configuration file should be
> kept until the configuration file for the next version is created.
> 
> The reason is simple.  The present strategy breaks `git blame`.  Take a
> look at what `git blame origin/master -- target/linux/x86/config-5.15`
> produces.  According to that, everything was created at 7ef75445c57.
> While this is strictly true, what one really wants is to know the history
> of the individual lines.

I shouldn't have focused on `git blame` here.  In truth this breaks many
git operations.  `git rebase` now needs a great deal of manual
intervention to get results.  I suspect `git merge` would also have
problems.

While this worked reasonably well for Subversion, for `git` this works
very badly.  Could development practices be adjusted to not break the
main tools which are in use?


-- 
(\___(\___(\______          --=> 8-) EHM <=--          ______/)___/)___/)
 \BS (    |         ehem+sigmsg at m5p.com  PGP 87145445         |    )   /
  \_CS\   |  _____  -O #include <stddisclaimer.h> O-   _____  |   /  _/
8A19\___\_|_/58D2 7E3D DDF4 7BA6 <-PGP-> 41D1 B375 37D0 8714\_|_/___/5445