[FFmpeg-devel] Who is against GIT now? (Was: [RFC] libswscale into the FFmpeg SVN repo)

Reimar Döffinger Reimar.Doeffinger
Mon Apr 6 15:58:41 CEST 2009


On Mon, Apr 06, 2009 at 02:18:07PM +0100, M?ns Rullg?rd wrote:
> Felipe Contreras <felipe.contreras at gmail.com> writes:
> > On Mon, Apr 6, 2009 at 4:07 PM, Diego Biurrun <diego at biurrun.de> wrote:
> >> On Mon, Apr 06, 2009 at 03:40:22PM +0300, Felipe Contreras wrote:
> >>> If you want to do non-standard things you need your own scripts.
> >>> Strangely the rest of the world is doing fine with SHA-1 id's.
> >>
> >> Strangely, some of the other distributed revision control systems do
> >> implement version numbers. ?At least mercurial and bazaar IIRC, which in
> >> practice amounts to everyone except git.
> >
> > By "rest of the world" I meant "the rest of the world that uses git".
> > But in any case, mercurial+bazaar are the minority here; most people
> > use git.
> 
> Bitkeeper uses revision numbers, and it's a royal mess when merging
> branches.  I've enver used hg or bazaar, nor would I ever consider
> doing so.

Well, I am not sure that there is a need for revision numbers, but
particularly for FATE the "necessary" thing IMO is a way to uniquely
identify a specific version that
1) allows ordering
2) allows to know if there were any changes in-between

I guess 1) could be solved by using
year-month-day-hour-minute-second
as one big number, 2) is a bit harder though you could
append the "revision" of the next version or something.
Wasn't some revision control system supposed to use fractional revision
numbers which would break only one of the two revision number systems
when merging (actually it could be used to keep all "mostly" intact by
e.g. merging revisons 542 and 934 into 542.934 etc.)?



More information about the ffmpeg-devel mailing list