Re: comment on patch-2.5.4

[Andreas Gruenbacher]

On Friday, 6 February 2009 19:19:07 Support Team wrote:
> Consider a patch file that patches the same file more than once
> in the sense that it contains more than one  output from "diff".
> (For example the Red Hat 9 kernel patch patches "kernel/signal.c"
> three times with three different runs from diff.)
>
> If numbered backups are made, applying such a patch causes one
> numbered backup for each diff, so it is possible to see what has changed
> at each stage.  However there is no versioning on the ".rej" file, so if
> there are rejects at more than one stage, only the latest set is kept.
>
> Yes, therre are workarounds, but also allowing numbering of the reject
> files would be nice in this case.

Interesting. A similar problem exists with backup files: patch doesn't 
remember which backup files it created, and for patches that touch the same 
file more than once, backup files can get overwritten.

The next version will have this fixed though: please see the "testing" branch 
of the git repository at https://savannah.gnu.org/projects/patch, or:

  git clone git://git.savannah.gnu.org/patch.git
  cd patch
  git checkout testing
  ./autogen.sh
  ./configure
  make

Multiple versioned reject files don't seem to make much sense to me: wouldn't 
it make more sense to put all rejects into the same reject file? (The 
infrastructure for implementing that is already there for the backup files.)

Thanks,
Andreas