tracing with runtime/trace
If sqlite.Pool moves to sqlitex as described in #47, it can take a context and that opens up the possibility of using runtime/trace for tracing.
As the sqlite package cannot depend on context, the trace object will have to be passed in from sqlitex using an adapter. That's easy enough.
A potential problem: what does tracing mean for an sqlite query? Commands like INSERT or UPDATE or SELECTs with a single call to sqlite3_step are easy, but multi-row queries are processed incrementally, so accounting is tricky.
Why can the sqlite package not depend on context?
In another issue I realized that you are avoiding context to avoid importing reflect for ease of use with tinygo.
Is tracing implemented in sqlitex at this point?