--- urn: "urn:penemure:8ce5a610-c44a-4012-8557-be8738ec7136" backend: "pub" ancestors: - 📁 Penemure Development Blog created: 2025-01-31 15:33:42.595176+00:00 updated: 2025-02-10 12:52:55.805219+00:00 tags: --- # 📓 SQLish: again In the beginning we had [`sqlish.go`](https://github.com/hexylena/project-management/blob/2d1d57074513ce4f81d0cc714c209baf6069db52/sqlish/sql.go) which was truly a horror and failed to implement anything useful of the sql standard. Next we had [`sqlglot`](https://github.com/tobymao/sqlglot) which was fantastic but, dear goodness it was slow. ``` $ SQLITE=false python scripts/sql.py "GROUP select type from __all__ group by type limit 1" Created HASH in 1.811981201171875e-05

 Executed query in 0.1616196632385254

 ``` wow! that's... terrible! Let's replace it with sqlite in-memory: ``` $ SQLITE=true python scripts/sql.py "GROUP select type from __all__ group by type limit 1"
 Created SQLITE3 DB in 0.006751537322998047 
Executed query in 2.7418136596679688e-05
 ``` oh. yeah. that's more like it. It takes far longer to create the database, but, it's far, far, *far* faster to query which is worth a lot.