2015-07-03 15:27:46 +0000 <saulzar_> You could have a situation with a local event loop which is going off continuously (large magnitude) which dominates
2015-07-03 15:27:47 +0000 <joco42_> https://usercontent.irccloud-cdn.com/file/pw5FWXl4/20150703_182729.jpg
2015-07-03 15:28:10 +0000 <joco42_> so this one in the center is a global node
2015-07-03 15:28:11 +0000 <saulzar_> Maybe it'd have to consider how far the data has come from the source
2015-07-03 15:29:47 +0000 <joco42_> https://usercontent.irccloud-cdn.com/file/CLJzUo3f/20150703_182947.jpg
2015-07-03 15:30:06 +0000 <joco42_> the nodes in the bottom circle are all local
2015-07-03 15:31:15 +0000 <joco42_> and the pagerank alogorithm would say the same of course
2015-07-03 15:31:26 +0000 <saulzar_> Yeah
2015-07-03 15:31:41 +0000jewel(~jewel@105-236-128-167.access.mtnbusiness.co.za)
2015-07-03 15:32:21 +0000latk(~textual@global-1-27.nat.csx.cam.ac.uk) (Quit: My Mac has gone to sleep. ZZZzzz…)
2015-07-03 15:33:20 +0000 <joco42_> another interesting measure is the histogram of pagerank numbers
2015-07-03 15:33:40 +0000 <joco42_> how many nodes have rank n ?
2015-07-03 15:33:48 +0000 <joco42_> so to speak
2015-07-03 15:34:49 +0000 <saulzar_> Would be kind of nifty if you could plug a large application and measure it fully automatically
2015-07-03 15:37:59 +0000 <joco42_> for traditional haskell (not frp) programs this approach is not very interesting because there are no (large) loops
2015-07-03 15:38:20 +0000 <joco42_> the interesting part comes from the loops
2015-07-03 15:38:42 +0000 <joco42_> i think
2015-07-03 15:40:11 +0000 <ryantrinkle> one thing that would be hugely helpful is to have some stack trace capability
2015-07-03 15:40:25 +0000 <ryantrinkle> primarily for loops, but also for other stuff
2015-07-03 15:40:39 +0000 <ryantrinkle> for loops, what you really want is a stack trace for each *node* in the loop
2015-07-03 15:40:53 +0000 <ryantrinkle> the stack trace of the loop exception itself is generally not helpful
2015-07-03 15:41:14 +0000 <joco42_> what would this stack trace contain?
2015-07-03 15:41:32 +0000 <ryantrinkle> the stack trace of each combinator that forms the loop
2015-07-03 15:41:40 +0000 <ryantrinkle> so, e.g., if i write
2015-07-03 15:42:12 +0000 <ryantrinkle> let e = mconcat [ fmap succ e, x ]
2015-07-03 15:42:22 +0000 <ryantrinkle> i should get two stack traces, one on the mconcat and one on the fmap
2015-07-03 15:42:38 +0000 <saulzar_> A tree of where the event came from?
2015-07-03 15:42:58 +0000 <ryantrinkle> yeah
2015-07-03 15:43:20 +0000 <ryantrinkle> i think it might be helpful to be able to focus down to just the loop, but the whole tree would be very helpful, too
2015-07-03 16:23:37 +0000BartAdv(uid90451@gateway/web/irccloud.com/x-epdglpvjulvcprlt) (Quit: Connection closed for inactivity)
2015-07-03 16:29:12 +0000bennofs(~benno@ipb2186cc7.dynamic.kabel-deutschland.de) (Ping timeout: 255 seconds)
2015-07-03 16:35:35 +0000latk(~textual@97e64c1f.skybroadband.com)
2015-07-03 16:35:36 +0000saulzar_(~oliver@121-74-230-192.telstraclear.net) (Ping timeout: 256 seconds)
2015-07-03 17:12:48 +0000ocharles_(sid30093@gateway/web/irccloud.com/x-bfpkgirtsanmfuxq) ()
2015-07-03 17:12:53 +0000ocharles__(sid30093@gateway/web/irccloud.com/x-xbhxjkbgdyvyxjhr)