Making the Blender C core extensible is on the scale of the 2.5 project, multiple years of work by many developers. Opening that discussion here is not going to help us make a decision on Collada now.
There was a thread about plugins recently on this mailing list, it's better to continue the discussion there. On Nov 27, 2015 08:40, "homac" <[email protected]> wrote: > Hey there! > > > > Not sure if I'm free to express my opinion here, but I feel somehow > related to this discussion due to my current project. > > Long story short: I suggest a radical change and move further towards a > pure plugin architecture (even on C language level), removing (i.e. > outsourcing) everything that's not blender core functionality into a > separate plugin project (no longer part of blender's source code tree). > > Blender has grown to a significant large system trying to fulfill all > kinds of user demands. Unfortunately, the development/maintenance effort > grows exponentially with the size of a software system due to the > interfaces between modules and necessary negotiations between the > different teams maintaining it. And people tend to stick to what they > already have which results in interfaces getting more and more messed up > with every compromise made to suite the needs of others due to the tight > integration. > > Consequently moving to a plugin architecture concept (again: on C > level), and removing everything not actually part of blenders core > functionality, cleans up interfaces and gives you more freedom for > future developments. The plugin teams on the other side have their own > freedom to write whatever they think is necessary, and they know exactly > what they can expect from core. (so, gaia: you could keep maintaining > collada if you so desire - that's why I wrote this mail ;) ). Due to the > same reason, it will attract more coders to contribute and the blender > ecosystem will grow even faster without you having to worry about it. > And blender can need it, because development seems a bit behind schedule > considering modern methods and technologies - as I realised in less than > a month working with blender core and listening what you are talking > about on channels. > > For those, who didn't get the picture yet: I think of Eclipse or Apache > server as some prominent successful examples. There is of course a lot > of ranting between the different groups, and plugins/modules tend to > have compatibility issues but the benefit is obvious, I believe. > > > > Regards > Holger > > > On 26/11/15 23:09, Brecht Van Lommel wrote: > > I think the main issue is that every feature in Blender has some > > maintenance cost. Refactoring, updating for new features, bug fixing, > > building library dependencies, etc, also for people who are not > > maintaining the module itself. Especially with big refactoring like > > Blender 2.8. Imagine for example big changes to the material system > > that would require Collada material I/O to be rewritten. > > > > The question is then if it's worth the cost. A few years ago Collada > > seemed like it could become a real 3D interchange standard, but it > > seems that hasn't happened. Looking ahead 2 years for the Blender 2.8 > > release, will it still be relevant then? > > > > Still there are some users, so it would be interesting to understand > > what they use it for. I've seen some people use Collada for basic mesh > > export from Blender into some smaller or custom game engines, but > > other file formats could work for that as well. For materials, > > animation, armatures I'm under the impression that I/O was never > > working well enough (both on Blender's and other application's sides). > > > > But maybe there will still be important use cases for it in 2 years > > and I'm just unaware of them. > > > > On Thu, Nov 26, 2015 at 9:45 PM, Gaia Clary > > <[email protected]> wrote: > >> Hello; > >> > >> Recently there was a serious discussion about removing the Collada > >> module from Blender. As far as i understand there is no decision > >> made yet, but i have the impression that all main developers want > >> this module to be removed. > >> > >> * The main problem with keeping Collada seem to be the size of its > >> libraries (some 40 MBytes) which need to be shipped with Blender > >> releases. > >> > >> * The main problem with creating a Python based alternative is that > >> Creating a Collada Importer for Blender is a challenge. Collada > >> exporters are not such a big problem. > >> > >> Can we get a decision on that? Depending on that decision i will > >> either continue to maintain the Collada module or let it go. > >> > >> But Please do not misunderstand me. If the decision is made to keep > >> the Collada module, then i will of course continue working on it. > >> Otherwise i will step to new challenges :) > >> > >> thanks > >> cheers, > >> Gaia > >> _______________________________________________ > >> Bf-committers mailing list > >> [email protected] > >> http://lists.blender.org/mailman/listinfo/bf-committers > > _______________________________________________ > > Bf-committers mailing list > > [email protected] > > http://lists.blender.org/mailman/listinfo/bf-committers > > > > _______________________________________________ > Bf-committers mailing list > [email protected] > http://lists.blender.org/mailman/listinfo/bf-committers > _______________________________________________ Bf-committers mailing list [email protected] http://lists.blender.org/mailman/listinfo/bf-committers
