On Thu, Jun 10, 2021 at 7:46 PM David Bremner <david@tethera.net> wrote: > > Felipe Contreras <felipe.contreras@gmail.com> writes: > > > We basically steal all the objects from their notmuch parents, therefore > > they are completely under Ruby's gc control. > > > > The order at which these objects are freed does not matter any more, > > because destroying the database does not destroy all the children > > objects, since they belong to Ruby now. > > I guess from a purist point of view this is a kind of layering > violation, since the use of talloc is purportedly an internal > implementation detail of the library. Still, I think it's a reasonable > approach given that the ruby bindings are maintained as part of notmuch, > and we are not very likely to abandon talloc. Yes, some refcount API could be added to all the notmuch objects, but at least I couldn't get talloc refcount stuff work for Ruby, so more work would be needed for that. > I am OK with applying the series as is. The only sensible thing I can > think of at the moment for testing is to run something like the script > of id:20210517193915.1220114-1-felipe.contreras@gmail.com as a "time > test", so not attempt to get valgrind working, but just run it on some > decent size corpuses and see that it it does not crash or leak too much > memory to complete. I have the test ready, the only question is how many times to run it in a loop. 100 times takes about 10 minutes with the large corpus size. -- Felipe Contreras _______________________________________________ notmuch mailing list -- notmuch@notmuchmail.org To unsubscribe send an email to notmuch-leave@notmuchmail.org