My approach would be to categorize user content into
- CRDT-enabled content considered as the norm, and
- content trapped in legacy containers that do not support CRDT.
With an implied warning for the user: legacy content offers a limited user experience.
Migration from "legacy" to "normal" should be fully supported, and as transparent as possible (but needs dedicated support for each file format). Converting "normal" (CRDT-enabled) content into files/blobs should be treated as a "take an snapshot" export operation.
I don't know. I guess they simply don't.
My approach would be to categorize user content into
- CRDT-enabled content considered as the norm, and
- content trapped in legacy containers that do not support CRDT.
With an implied warning for the user: legacy content offers a limited user experience.
Migration from "legacy" to "normal" should be fully supported, and as transparent as possible (but needs dedicated support for each file format). Converting "normal" (CRDT-enabled) content into files/blobs should be treated as a "take an snapshot" export operation.