Changes between Version 2 and Version 3 of Ticket #34255, comment 17


Ignore:
Timestamp:
Jan 26, 2023, 7:38:46 PM (22 months ago)
Author:
Simon Charette

Legend:

Unmodified
Added
Removed
Modified

  • Ticket #34255, comment 17

    v2 v3  
    11I agree with Tim, having a bit more documentation on the known limitations/bugs of server-side binding (we should probably create an issue to track the group by one or maybe use #34262 with a dual purpose) and [https://medium.com/engineering-at-birdie/puzzling-postgres-a-story-of-solving-an-unreproducible-performance-issue-778075ed7998 its gotchas] which have ties to the usage of prepared statements (#20516).
    22
    3 FWIW I was surprised to learn that `psycopg>=3` also [https://www.psycopg.org/psycopg3/docs/advanced/prepare.html#prepared-statements prepares statement automatically on the fifth execution] which might result in unexpected behaviour. I'm not sure if this is the case when client-side binding is used though, I assume not as that requires the usage of server-side bindings?
     3FWIW I was surprised to learn that `psycopg>=3` also [https://www.psycopg.org/psycopg3/docs/advanced/prepare.html#prepared-statements prepares statement automatically on the fifth execution] which might result in unexpected behaviour.
     4
     5~~I'm not sure if this is the case when client-side binding is used though, I assume not as that requires the usage of server-side bindings?~~ confirmed from [https://www.psycopg.org/psycopg3/docs/advanced/cursors.html#client-side-binding-cursors the docs]
     6> Client-side cursors don’t support binary parameters and return values and **don’t support prepared statements**.
    47
    58All that to say that we should probably have a section with a summary of what to expect when switching to `psycopg>=3` and enabling server-side binding and maybe advertise the support as ''experimental'' for this first release?
Back to Top