--- Log opened Sat May 02 00:00:02 2009
16:23 < mae> working on the lock system..
17:30 < mae> stepcut i just pushed up some lockfile updates, can you please test them? :)
20:53 < stepcut> the comments about FromReqURI are wrong. Is it too late to change them for 0.3?
20:54 < stepcut> it says that fromReqURI is exactly readM. But that is not true. It behaves different from String
20:55 < stepcut> *Blog.State Happstack.Util.Common Happstack.Server> readM "foo" :: Maybe String
20:55 < stepcut> Nothing
20:55 < stepcut> *Blog.State Happstack.Util.Common Happstack.Server> fromReqURI "foo" :: Maybe String
20:55 < stepcut> Just "foo"
20:55 < stepcut> *Blog.State Happstack.Util.Common Happstack.Server>
20:59 < stepcut> the comment for 'path' is all bitchy about it too. But, practically speaking in is nice that you can use path to extract /somestring/ instead of /"somestring"/
22:03 < gcollins> i hit some mega-weird performance issue with happstack yesterday
22:03 < gcollins> i had a happstack server waiting on http requests, and a little haskell fuzz tester sending client messages
22:04 < gcollins> to improve throughput i was pipelining the requests --- and the server started spending 99% of its time inside "futex"
22:05 < gcollins> i turned pipelining off, and cpu use went back to normal
22:06 < gcollins> i suspect the problem might be in the happstack network code but it could also be in the ghc runtime system -- i don't really have any idea how to start tracking this one down, if anyone has any ideas please let me know
22:46 < stepcut> gcollins: did you compile with -O2 and -threaded?
--- Log closed Sun May 03 00:00:03 2009