Fork me on GitHub

@alexmiller based on a discussion in #clojure I took another look at and decided their root cause analysis didn't go far enough, so their conclusion (they need a new feature, non-shared timeouts) was faulty. I dropped a patch on the ticket that I think solves their real issue (the non-selected channel operations in an alt retain a reference to closed over data potentially indefinitely), without adding a new feature. The ticket however is still tagged as a 'new feature' and 'minor', so I wanted to ping you because I think my patch solves their root issue, doesn't add a new feature, and has a large potential impact on memory usage in core-async heavy programs.

😻 3
Alex Miller (Clojure team)19:12:24

maybe a new ticket would be better?

✔️ 3
hiredman20:12:01 is the new ticket, I'll move the patch over, and see if I can get some graphs of differences in memory use pre/post patch

Alex Miller (Clojure team)20:12:00

thanks, that is a good writeup