hubert@feyrer.de wrote:

> IMHO we should not do this.
> 
> Moving files around is one thing that'll bear enough errors, collecting
> patches is another one. For me, I'd still prefer to have the patches
> seperated, and not being tied to some tools.

  I have always been a fan of separated patch files.  But if the goal is
to reduce inodes, we definitely need to do something about the large
number of patch files.  IMHO, having tools that split and join patch
files as part of the package system is a good compromise between ease of
handling and reduction of inodes.

  To avoid any oversights on our (the developer's) parts, these tools
need to be part of the package building process.  I.e., never apply the
huge super-patch directly, but split it up into (e.g.)
${WRKDIR}/.patches/patch-* first, then apply the individual patches. 
This way, the only inconvenience for us developers is that we need (a
trivial) script that concatenates ${WRKDIR}/.patches/patch-* back into
one patch file.

  This is IMHO no more error-prone than the current procedure of
updating the individual patch files in the patch subdirectory and
running 'make makepatchsum' afterwards (especially if the latter is
automatically handled by the re-integration script).

  Cheers
      ,
   Rene