@greg This failure keeps happening every few days on DrDr: http://drdr.racket-lang.org/40052/racket/share/pkgs/http/http/request.rkt

is there a way that test could be made more robust?

IME the answer to “how can HTTP connections be made more robust?” is often “retry with exponential backoff”. I can add that. Although may just hit the test timeout, instead. Let me think about it ….

(Sorry about the noisy test)

Does anyone know how this Redex Hackfest went? (See https://groups.google.com/forum/#!msg/racket-users/min-CDue3Bc/MDwze5Ijqp0J and https://github.com/racket/racket/wiki/Redex-Features)

In particular, I’m interested in the “disable caching for individual metafunctions” feature, which I can’t find in the documentation.

greg: I’ve got a package for that FYI :)

@notjack Cool. Which one?

greg: retry

:slightly_smiling_face: OK thanks I’ll check it out.

Does connection-pool-lease from the db module return connections when a thread is killed with kill-thread? The docs seem to imply so, but they just say “when thread is terminated” and I’m not sure if that’s limited to the thread only exiting cleanly.