I tried. It's not easy because of how limiting SQLites indexes are. You have to build your own indexes using `TRIGGER`s or in a software wrapper and tables.
Commendable attempt! I've considered writing a datalog storage backend on sqlite just like your prototype. Thank you for sharing, now I can lazily study your prototype instead of doing the hard work myself. :) I'm curious, what kinds of limitations of SQLite indexes are you referring to?
Sparse indexes are pretty limited and it only supports B-tree, which make implementing AVET and VAET difficult. Further efficiently finding the current value for a E + A is difficult to do in SQL in a way that doesn't require maintaining a whole copy of the data. I actually bumped up against what I believe are weird edge-case bugs in the SQLite query planner when dealing with sparse indexes as well.
I think I gave up when trying to implement one-many relationships because the SQL was getting too gnarly.
