#haskell - Sun 4 Mar 2007 between 18:39 and 18:46

NY Lost Funds



Philippasorear: I believe so
matthew-_sorear: I've certainly left Epigram running overnight ;-)
sorearwhat makes nonterminating compiles worse than arbitrarily slow ones?
david_nmessenger : it's the way the array is made that makes my neuron burn.
actionsorear pragmatically equates "forever" with "until the heat death of the universe"
sorearpragmatically equates "forever" with "until the heat death of the universe"
david_notice how I put neuron at the singular :D
augustsssyntaxfree: give me 10 minutes in front of a whiteboard and you'll grok existential types
nmessengerarray bounds (zip indeces elemnents)
SamBsorear: I'd say that arbitrarily slow ones are worse when you believe that nonterminating ones are possible
matthew-_yeah, but you just have this feeling that if you leave it running for 1 more minute it'll termiate
Philippathat slips from "fast enough" to "never" are more common than slips from "fast enough" to "hello Ackerman!"
SamB(because you'd tend to think you had a nonterminating one)
matthew-_mmm. maybe non-terminating is better because if it's still running after a day you might give up and rework the problem from a different angle
with terminating, you're likely to sit there stubbornly until it terminates
SamBhmm.
augustssnon-termination is detectable :)
yipthis is what progress bars are for =)
SamBwell, actually, I'd probably do it anyway in case I was maybe using an exponential algorithm
matthew-_augustss: in finite memory, sure. Or more generally?
SamByip: progress bars require analyses of how the algorithm progresses
augustssmatthew-_: in finite memory. do you have more than that?
sorearmaybe epigram should be modified to non-optionally abort compilation after 10 minutes :)
matthew-_SamB: not on Windows they don't!
sorear: I suspect you can do that with alarm
actionallbery_b just saw something about how progress bars are evil because they rarely model what actually takes what amount of time
allbery_bjust saw something about how progress bars are evil because they rarely model what actually takes what amount of time
SamBmatthew-_: now, you could do one of those "still kicking" bars
matthew-_augustss: yes, I have 3 times that much! ;-)
Korollaryprogress bars are useless.
SamBbut of course a nonterminating algorithm could be kicking too
augustssmatthew-_: then it'll just take slightly longer to detect
allbery_b(case in point: install jdk with documentation on osx :)
matthew-_augustss: where "slightly" is some exponential function, I'd guess?
sorear@quote matthew-_ <augustss> matthew-_: in finite memory. do you have more than that? <*matthew-_> augustss: yes, I have 3 times that much! ;-)
lambdabotNo quotes match. Your mind just hasn't been the same since the electro-shock, has it?
sorear@remember matthew-_ <augustss> matthew-_: in finite memory. do you have more than that? <*matthew-_> augustss: yes, I have 3 times that much! ;-)
lambdabotDone.
augustssmatthew-_: yes
matthew-_cheers sorear ;-)
actionSamB wonders if Lemmih's GDB is buggy
SamBwonders if Lemmih's GDB is buggy

Page: 2 9 16 23 30 37 44 51 58 65 72 79 86 93 100 107 114 121 

IrcArchive

NY Lost Funds