On Tue, Jan 2, 2024 at 12:31 PM Nathan Bossart <nathandbossart@gmail.com> wrote:
> I think we're supposed to omit the "Lock" suffix in wait_event_names.txt.
Ugh, sorry. But also, why in the world?
> > It seems like it would be good if there were an automated cross-check
> > between lwlocknames.txt and wait_event_names.txt.
>
> +1. Here's a hastily-thrown-together patch for that. I basically copied
> 003_check_guc.pl and adjusted it for this purpose. This test only checks
> that everything in lwlocknames.txt has a matching entry in
> wait_event_names.txt. It doesn't check that everything in the predefined
> LWLock section of wait_event_names.txt has an entry in lwlocknames.txt.
> AFAICT that would be a little more difficult because you can't distinguish
> between the two in pg_wait_events.
>
> Even with this test, I worry that we could easily forget to add entries in
> wait_event_names.txt for the non-predefined locks, but I don't presently
> have a proposal for how to prevent that.
It certainly seems better to check what we can than to check nothing.
Suggestions:
- Check in both directions instead of just one?
- Verify ordering?
--
Robert Haas
EDB: http://www.enterprisedb.com