That’s not why we’re moving it. We’re moving it because it’s not maintained and has some questionable licensing. (We had to remove one file recently so that JME doesn’t have to convert itself to GPL… who knows how many other files were mislicensed from the beginning.) When we move it to its own repo, it can be GPL if it likes.
Also, if it was maintained then I’m sure the person maintaining it would probably like to release it more often than JME releases.
The .blend format has always been super buggy. You had to make your models in a pretty specific way or random problems would crop up. It’s not an game file interchange format and so will always suffer from these issues… it has to support the entirety of Blender’s features (not just the small percentage useful for games) and has no other master than Blender itself. They could completely change the format 100% tomorrow and it would not affect Blender. There is no real motivation for them to not break .blend loaders because the format is not even meant for that.
GLTF on the other hand is specifically designed for this purpose and has industry backing. The GLTF exporter is open source and so any problems found can get fixed just like JME’s .blend importer… the difference is that there are actually folks working on that.
The idea of an exporter is better in the first place because it’s better suited to “dumb down” all of the Blender features into something “game ready”. It’s sitting right in Blender, after all.
The alternative of “support all of Blender” in an importer (which is what the JME Blender tries and fails to do) is sort of doomed from the start. It will always miss something and always be 9 steps behind even if properly supported.
I have never ever said blend import should be the main way to import models nor did I deny GLTF merits. I perfectly know Blend importer situation and I am ok with it which means using older Blender version and simpler models. Yeah GLTF works for everyone but it is not a friendly way to export models especially for noobs. Blend importer covers small but not insignificant number of modelling cases. No reason to get rid of it especially if the main purpose of current SDK existence is teaching basics.
I will make this prediction, though: by this time next year, the only thing we will hear about the .blend importer is from users who mistakenly still try it and wonder why it doesn’t work with 2.8. (Regardless of where it lives.) We’ll direct them to just use the built in GLTF exporter in Blender and that will be essentially the end of those threads.
(Note: when we do direct them to the GLTF exporter, we should mention to use the gltf format and not the glb format… it’s really the only hits-everyone-once ‘gotcha’ in that process.)