No repro yet, but putting this here for further investigation.
I had a query that was taking 30 seconds to load via the Python supabase client(uses PostgREST), then I switched to psycopg and it brought it down to 5 seconds.
it was just a select all from a table with 151,898 rows and each row has 60 columns.
No repro yet, but putting this here for further investigation.
(From reddit)
I've seen other cases where wide tables(with a good amount of columns) results in slow responses.
For now I believe this could be related to Hasql decoding.
Related: https://github.com/PostgREST/postgrest/discussions/2831