@danielcompton I looked at this more carefully just now. Looks like there are already ongoing discussions about handling auth, but no complete packaged solution. Still missing are: - Wrapping this up into a Vase sample - A doc comparing Friend vs Buddy vs Geheimtur I don't know quite enough to do either of these tasks myself, but I'm wrestling with whether I should I extend my current bikeshedding to learn enough. For my current toy project, the auth in https://github.com/cognitect-labs/vase/blob/master/samples/petstore-full/src/petstore_full/interceptors.clj is really all I need, so my rational side is arguing to not waste more time trying to do it right. For future reference: One conversation (@ddeaguiar and @ohpauleez) leading to PR for a demo in Pedestal: https://clojurians-log.clojureverse.org/pedestal/2017-03-03.html, https://groups.google.com/forum/#!topic/pedestal-users/MJL0h4ts_mk; https://github.com/pedestal/pedestal/issues/500, https://github.com/pedestal/pedestal/pull/503 An existing Pedestal-based sample (@propan): https://github.com/propan/geheimtur-demo

Hi. Was the Pedestal entry in the Techempower Web Framework Benchmark now with good production settings? New results are out, much better but far from good (all clojure entries imo): https://www.techempower.com/benchmarks/diffs/r13-r14.html & https://www.techempower.com/benchmarks/#section=data-r14&hw=ph&test=fortune

@ikitommi I’ve passed your question on to Paul who was involved in the benchmarks. Will post here when I find out.

(My commentary: I don’t understand the TE benchmark… I’ve seen 200K responses per second on my local hardware. But that is with routes designed to use the map-tree router which is super-fast.)

@mtnygard thanks

FWIW, Paul also says that the published results don’t match any configuration we’ve seen. Way lower on all ratios.

That’s across loopback, physical gigE connections, and AWS