It’s good to take into account blender 4.5 which still has collada support is an LTS release, so it will be supported well into 2027. If your community relies heavily on the collada format you should take this thread as the opening shot for moving to a different format (usd and gltf could both be viable replacements) 2 years is a long time. If you just sit and do nothing until the clock runs out you really can’t blame blender anymore, there was plenty of time to do something.
with that said, it’s not like the blender versions that do support collada are going anywhere, blender 1.80a (25 years old!) still starts on my machine. You can always keep a copy of 4.5 around for import/export purposes.
While it may not be his business what you do on your own copy of the software on your own computer, what you claim to want to do with Blender on the official forum for Blender development discussion is a concern for anyone discussing development.
General rule of thumb anywhere: if you are doing questionably legal things, don’t talk about it in an official forum. If you do, expect someone on that forum to decry your actions. Otherwise it might be taken as condoning them, with all the potential legal problems that come with it.
I’m pretty sure the devs do not want that sort of headache. Other users don’t want the project suffering from those sorts of issues either. While the likelihood of you specifically bringing all that to Blender’s doorstep is low, it’s never zero, and even if it was allowing mention of dodgy actions to go unchallenged could damage Blender’s reputation.
So no, he’s not Nintendo. He just doesn’t want Nintendo to come knocking on the door of software he relies on because of someone else’s indiscretions.
Apologies for reviving this old post. I’m not a dev by any means I just wanted to loop back to answer the original comment of this thread. We do have glTF support at Second Life (Linden Lab - I’m not an employee there just a user of the platform) but only for PBR materials as of right now and we are still solely reliant on the .DAE format to upload mesh models. I don’t know what’s in the pipeline in terms of expanding glTF support for mesh models, but we don’t have that option. This has been worrying to some of us in the community that the support may be ending and has been moved into a legacy state. I don’t understand the nuances of a decision like this I just hope my information helps answer some questions earlier in the thread.
The way I see it there is the ideal programmer scenario (Collada is dead and everyone has/should just moved on) and the actual user scenario (the format exists out in the wild and is still being used albeit in less modern pipelines/asset libraries). I get the need to leave the library behind due to rot and bloated size but the remaining actual users who rely on it should be considered.
Perhaps the core support should be scrapped and the cleaned up version of the library can form the backend of an addon port of the importer/exporter hosted on the extensions platform? It would debloat and free blender from having to maintain a working build of the aging library and just leave building the occasional wheels for new platforms and keeping bpy updates to be the maintanence factor. The torch of support would be handed off to the community and not be a concern of core devs that way.
(Edit: actually it seems like pycollada is the better way forward in the extensions approach since no bindings seem to exist for opencollada.)
A while ago most of the worry was known security vulnerabilities in the dependencies that OpenCollada uses. The actual library being effectively dead and no one updating it for years. So I made my own fork that at least updated the dependencies to the latest versions with no 60+ security vulnerabilities (GitHub - aras-p/OpenCOLLADA). Now Blender uses that version.
But yes, besides that there is also library size (OpenCollada, due to extremely over-engineered way it is written, contributes quite a lot to blender build size). And just general maintenance.
Yes, that would be ideal. Collada importer/exporter written as a Blender extension. Then there’s no maintenance burden on regular Blender developers (for arguably a nice, and decreasing importance, format), no blender size overhead for “literally everyone”, etc.
The only question is, is the community of people who want/need Collada large and determined enough to write and maintain such an extension. I don’t know the answer.
Given the much more obscure, oddball, and outdated formats that have importers/exporters on the extensions platform already (eg. who needs to import/export meshes/scenes from UE1 in 2024?) my hunch is that there would absolutely be new set of community maintainers of the importer/exporter fairly quickly given how common it is to stumble upon dae assets/old software using it out in the wild.
Should the support survive long after the extension hand off or not either way it would be a proper send off compromise. In the end it would die a natural death of irrelevance (in a much more flexible and revivable state) rather than be taken behind the shed for being a burden.
I think that would be wrong. The time to start making a python addon was last May when this thread started and yet there’s been nobody. The current collada support is written in c++, not python, so there won’t be a “just move this to an extension” since that’s not possible without a significant effort.
The community has always been free to make their own. They didn’t. Linden Labs staked their entire foundation on collada. They haven’t contributed to collada. Other high profile games did the same. They haven’t contributed to collada.
Entity-A stakes their entire existence and foundation on Technology-B
Entity-A returns nothing to the Technology-B ecosystem nor provides any stewardship around it
Entity-A is somehow shocked when Technology-B goes away because there's no one maintaining it
Just want to point out that modding - sometimes very old games - is a thing.
But I agree with the general sentiment of dropping dead formats. If some community want certain I/O then this community probably should contribute towards it.
The motivation to write a new Python Collada extension, with its own new and exciting bugs and incompatibilities and Python-tier perf, and then maintain it forever, is obvious going to be very low until the existing Collada support is actually gone. And then the easiest way forward is, like has already been said, just to use an older Blender that still supports Collada.
In fact, the simplest way to make a Python importer right now is take advantage of Blender’s backwards compatibility and just shell out to an old Collada-supporting Blender which does the import, saves to a temp file, and then appends to the main file.
You could wrap that up in an operator, add all the import options, and have an importer in about 200 lines and a couple hours at just the cost of having to keep an old Blender hanging around somewhere. (And no more import crashes!)
