On Sun 20 Dec 2020 at 19:22 -0400, David Bremner wrote: Hi David, I'm just catching up on some mailing lists... > David Bremner <david@tethera.net> writes: > >> David Bremner <david@tethera.net> writes: >> >>> I hope Floris (or someone) can tell me what is going on here, I don't >>> understand the memory management in the CFFI bindings very well. >>> >>> As part of my attempt to remodel config handling, I've added a cache >>> for the configuration information stored in the database. This doesn't >>> seem to break any other tests, and valgrind doesn't spot any memory >>> errors. >>> >>> Here's the output from pytest >> >> Oh, I bet I know what it is. I need to update the cache when calling >> notmuch_database_set_config. It's just a surprise that nothing in our >> test suite does a read after write for configuration information in the >> same database session. >> > > Yep, that was it, sorry for the noise. I guess I kindof panicked at not > understanding what was going on in the bindings tests. Yes, this test is checking deletion actually works. Glad the python tests helped catch an issue. :) Cheers, Floris _______________________________________________ notmuch mailing list -- notmuch@notmuchmail.org To unsubscribe send an email to notmuch-leave@notmuchmail.org