[Mplayer-cvslog] CVS: main/libmpeg2 mpeg2_internal.h,1.11,1.12 slice.c,1Wed Nov 14 23:01:37 2001

[Arpi]

Hi,

> 'cvs update' fscks up things when patches are reversed :/
> 
> Arpi: it should really be a new commit when you want to reverse patches:
> that way, a real history of changes is kept and the same patch is not
> reapplied again and again... (plus it fixes the mess with versions
> numbers)

removing a patch usually doesn't cause any problems, cvs client also
downgrade file if it see it. the problem come if the removed patch is
followed by a new commit with the same version number.
it never happened before this mpeg2_quant stuff. in such cases, the commit
following the reversed patch should change version with -r option...
so, for example:

v1.1  original
v1.2  bad patch, removed
v1.3  last commit, patch for v1.1 (cvs commit -r1.3)

so users who have teh bad v1.2 patch will get v1.3 too.
(or maybe their client compains about cvs merge conflict, if the client only
downloads the patch between v1.2-v1.3... but it shouldn't happen as patces
are generated by the server)


A'rpi / Astral & ESP-team

--
mailto:arpi at thot.banki.hu
http://esp-team.scene.hu