[Bf-committers] Removing Build Systems

Shaul Kedem shaul.kedem at gmail.com
Wed Apr 15 15:09:45 CEST 2009


Hi,
 Although I love cmake, I think we should stick to both cmake and
scons; for example, I tried to get python to have a postfix of debug
("_d") in msvc and it turns out that cmake's support (or
documentation, or community, or whatever) is lacking - I got it after
lots of tinkering and looking at the web, but I'm not sure that this
will be the case for everything in the future.

 Since most of us know python to some degree I think we should at
least have a way to always compile blender and that is scons on
supported platforms. I also think that anyone making a change should
be responsible to make scons and cmake work in the future (if that's
the decision)

Shaul

On Wed, Apr 15, 2009 at 4:35 AM, Nathan Letwory <jesterking at letwory.net> wrote:
> 2009/4/14 Brecht Van Lommel <brechtvanlommel at gmail.com>:
>> Beyond that I do think in general there is a significant amount of
>> time being wasted because we have this many build systems, but if
>> people want to invest time to keep their build system working I don't
>> want to stop them. It's just that in practice it seems to me that it
>> does not completely work this way and during development I still need
>> to deal with the other build systems anyway for various reasons.
>
> I still maintain SCons, the Blender 2.5 branch has some very good
> improvements already (which I don't want to backport to trunk, because
> I tried and it is just ugly work that will be phased out soon anyway).
> I'd certainly have some days of mourning if we decide to drop also
> SCons support, but if there really is a working solution, then I'm
> fine. I've tried CMake, but I find it still somewhat lacking in
> supporting everything we have - if someone manages to fix it and
> maintains it well, too, then I guess I have no further good arguments
> for keeping SCons (other than the amount of invested time and
> continuing work I do locally, waiting to be committed one day).
>
> There are three things that are important for me:
> * ease of use, meaning out of the box compiling
> * cross-platform - working on all our major platforms
> * flexibility (what you get with python for scripting galore)
>
> Of course, 1 is rather objective, but in the past months I tried using
> both the Make and CMake solutions, but only the CMake came close
> enough for the ease of use (which in turn maybe due to the fact I
> haven't used it really much). CMake and SCons both are good for
> cross-platform (Make is not) and from what I can see only SCons offers
> the flexibility that I'd like to have). The lacking part of SCons
> naturally is its speed as compared to Make when looking at rebuilds,
> but further I think it is a valid and working solution.
>
> In the end, I'm happy if any decision we make ensures the development
> potential in every sense now and in the future.
>
> /Nathan
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers
>


More information about the Bf-committers mailing list