Jean Boussier
Jean Boussier
😞, do you have any easy to understand literature you'd recommend on this topic? I'd like to better understand the constraints here. While I get the write are not atomic,...
Ok, I think I understand what you mean, and thank you for the link. If I understood the link correctly, given that the cache is only ever cleared from the...
Ok, thank you. So short of having 128bit atomics, or of packing that bit in the `VALUE`, there isn't a proper way to re-inline these caches 😢
Or I could inline them only if no other ractor has been spawned yet?
> you can treat all embedded caches as busts in multi-ractor mode and fill and clear the flag. I don't think you need to though. You can keep the embeded...
Interesting. Somehow the fix no longer works once backported :/ No idea to fix that.
This has been fixed years ago, you need to upgrade the gem to at least 0.5.4: https://github.com/brianmario/mysql2/pull/1083
cc @kddnewton nothing urgent, but perhaps this is something you understand better than me. With the old compiler, `keyword->table` seem to always points inside the `iseq->body->local_table`: - https://github.com/ruby/ruby/blob/436d23f80da20ab56ebe42da4f37fc5c3e276f69/compile.c#L1986 - https://github.com/ruby/ruby/blob/436d23f80da20ab56ebe42da4f37fc5c3e276f69/compile.c#L11794...
> Is the thread causing you any problems No, it's more that https://github.com/rails/rails/pull/52632 and https://github.com/rails/rails/pull/52622 made me thinking that 60 seconds (which likely nobody changes) is a very long time,...