Perhaps a controversial change? It seems preferable to me to punt this to the DB-API layer for implementation/optimization.
I don't think the SELECT hardcoded on the current implementation works on postgres (untested)... confusingly, this change may not help postgres support... but it doesn't break mysql nor sqlite.
Perhaps a controversial change? It seems preferable to me to punt this to the DB-API layer for implementation/optimization.
I don't think the SELECT hardcoded on the current implementation works on postgres (untested)... confusingly, this change may not help postgres support... but it doesn't break mysql nor sqlite.