SQLite Forum

CBOR Support & challenging DB design problem for keeping history with big schema changes
> was brought up a long time ago when JSON1 was introduced, and it's unlikely to come from the SQLite team itself, given the discussion at the time
> [...]
> and unlike the discussion at the time bring some speed-advantage over JSON1

I presume _a long time ago_ mentioned tag extensions did not exist yet, and it is what makes CBOR unique amongst all "binary JSON" variants, which are, in a nutshell, usually are not better _much_. So that discussion is most probably outdated, given that **reason is not speed** but **features**:

* compression
* ability to express things **impossible in JSON** (binary values, type info)
* extensibility (JSON itself and all other binary JSONs are _not_ extensible)

However, my main questions are **not about CBOR** support - if I could plug a function of my language's implementation into SQLite, this will be just fine.

The main question is how to build indexes on such, and AFAIK in JSON1 such indexes are also not possible.