Vahur Krouverk wrote:

> 
> With POV-Ray 3.5 release there is considerable interest in source code
> 'hacking', many users fix bugs and add new features. Perhaps it would be
> good to unite such efforts and create one version, which contains all
> useful modifications. 

Other projects have two versions: a stable and a developer version.
The first one is for people who actually want to use the program for
larger projects and therefore cannot afford to live with changing
syntax and unpredictable crashes. The second version is for people
(like me) who like to toy around with the newest, bleeding edge features 
and don't care too much, if the program can't process a file that was
created for a two weeks old version of the program. This approach is
successfully applied to several larger projects including the Linux
kernel, so my suggestion is to copy it.

As it comes to POV-Ray, the "stable" version is probably the official
release. AFAIK the POV-Team includes known bug fixes in minor updates.

For the "developer" version -- I think -- we have to distinguish between
two cases: versions that are based on 3.5 ("3.5 MegaPov") and the 
forthcoming 4.0, because the latter one is a rewrite from scratch. 
So, from my point of view, we need three versions of POV-Ray.

> Best would be, if there is public (CVS) server,
> which hosts source code and users can commit fixes and updates to it.

I'm against a cvs server where anyone can submit patches. Usually
you have some kind of maintainer who decides which patches are
suited for this particular version. In the before-mentioned scenario,
a cvs server doesn't make much sense for the official version.
There are just too few realeases.
Furthermore, it's up to the POV-Team to decide whether 4.0 will
be developed using a public (read-only) cvs server. (I would really
love to see this happen and therefore hope that rumors about it are
true.) 
Finally, for the 3.5 based MegaPov, a public cvs server would make
even more sense, because then developers could easily create their
patches for a common source base. Nevertheless I think that a
(group of) maintainer(s) is needed to ensure some kind of minimal
standard. If someone - as it has happend in the past - shows up
creating excellent patches I assume that such a person would get
write access to the cvs very quickly.

Just my Euro 0.02
Thomas

Post a reply to this message