Re: [Texmacs-dev] Monitoring changes in mainline

[david]

On Fri, Apr 18, 2003 at 07:27:33PM -0700, address@hidden wrote:
> 
> What is the rationale not to have the source in CVS at savannah.gnu.org?

CVS is not flexible enough.

Esp. it lacks important features like tracking moved file and atomic
commits of multiple files. Since TeXmacs is under heavy
reorganization, not having these features rules CVS out.

Oh, yes subversion has both... But (I assume) there is a perception
that still requires too much housekeeping.


> If only Joris has commit access and he makes a CVS tag each
> time he does a release I think there would not be much
> difference in his current procedures and there would be benefits
> both to him and to others without much cost.

Mhhh... yes that would not be a significant cost to Joris.


> CVS is not perfect, but in my experience it is far better than
> the model where the maintainer accepts patches sent in from
> others and does periodic releases. Even if only one person has
> commit access, CVS is better.
> 
> Even for the non-maintainer who has only anonymous access
> it is SO convenient to be able to 'cvs diff -u' and get a
> patch, or 'cvs update' and get a merge with the current CVS.

Then you hit some peculiarities of TeXmacs source tree. The version
number is hardcoded in may files (and a major directory name) so what
you propose is either plain unusuable (if the directory name issue is
not fixed) or will introduce a lot of noise (because of harcoded
version numbers).

If the version number issues are not fixed that is not usable.

If they are fixed, and the chosen system supports renames, that only
brings the practical benefit that I would not have to figure out how
files where shuffled during reorganizations.

Anyway, thanks for the suggestion.
--
                                                            -- DDAA