added a comment - - edited
I've been reviewing all this along the last days, and it's great code but have found some problems that, after a lot of thinking and sharing it with Martin have leaded to the next conclusions:
1) CC-Import: It's ok to send it upstream. It uses current support for backup converters in a proved way (1.9 => 2.x is using that already), integrates perfectly in the restore process and uses a lot of the logic that was present in the 1.9 importer. Also, it's somehow a bug-fix as far as restitutes one functionality we lost in 2.x.
2) CC-Export: This cannot be integrated right now. It is new code, leading to one integration with the backup process that needs to be planned/discussed/agreed a bit more. This is a (cool) new feature, but it's not ready for the masses and, right now, roughly 2-3 weeks to release, is not the time to fix it.
So, the immediate plan to get as much as possible of this landing is:
1) Create one new $CFG->experimentalccexport setting, without UI, only available manually @ config.php that will "enable" cc-export, so backup won't get any change by default. This is the optimal solution, so all the code will land already in core and we'll be able to work there after release and will enable devs and other sites to use it as experimental feature if wanted.
2) If I cannot isolate and get 1) done, I'll fallback to a more radical solution, deleting all the cc-export code interfering with "normal" backup. Then, after release, we can re-introduce it safely in future master branch (that will be 2.3 in 6 months).
And that's the plan for NOW. I'm working on 1) above right now aiming to release 2.2beta in hours.
Once again, I want to insist that I've liked the ideas/code I've seen, absolutely. Great job, guys!
PS: Side note, we would need some MDQLA tests covering CC-Import, do you've something about that (packages, tests...) we can incorporate to our MDLQA iterations?
PS2: While trying cc-import I detected that some resources were not imported properly, leading to "file not found" errors, it was using the well-known "OrganicChemistry.zip" 1.0 package. (seem to be file-related problems both in file resources and question images)