Am Mon, Dec 21, 2020 at 02:24:37PM +0000 schrieb Daniele Varrazzo:
> What I'm thinking about is to prepare queries automatically with a schema such:
>
> - decisions are made after the query is transformed to postgres format
> (i.e. it is reduced to bytes, all the client-side manipulations have
> been done, placeholders have been transformed to $ format). There is
> an object in psycopg3 that takes care of this transformation [2]
> - the number of times a query is seen is stored in a LRU cache on the connection
> - if a query is seen more than `connection.prepare_threshold` times
> (proposed default: 5) then it is prepared with the name
> f'pg3_{hash(query)}' and the following executions are prepared.
> - if more than `connection.prepared_number` queries are prepared, the
> one used least recently is deallocated and evicted from the cache
> (proposed default: 100).
> - Parameters may be fudged on the connection: prepared_threshold=0
> would prepare all queries, prepared_threshold=None would disable
> preparing.
> - For the control freak, cursor.execute(query, params, prepare=True)
> would prepare the query immediately, if it isn't already,
> prepare=False would avoid preparation. The default None would enable
> the automatic choice.
:-) seems like a well-balanced scheme to me !
Karsten
--
GPG 40BE 5B0E C98E 1713 AFA6 5BC0 3BEA AC80 7D4F C89B