January 2005 - Haskell-Cafe

RE: [Haskell-cafe] Question on "case x of g" when g is a function
by Bayley, Alistair 27 Jan '05

27 Jan '05

Another way of looking at it: case peforms pattern matching; it is _not_ an equality test. If you want equality tests, use if-then-else, or something like this (using guards): f x = case () of _ | x == bit0 -> 0 | x == bit1 -> 1 The behaviour that initially threw me was that case works for various literals (numbers and strings), but that's just pattern matching (see 3.17.2, item 7): http://www.haskell.org/onlinereport/exps.html#pattern-matching Another shorthand for f, using guards: f x | x == bit0 = 0 | x == bit1 = 1 > -----Original Message----- > From: Salvador Lucas [mailto:slucas@dsic.upv.es] > Sent: 27 January 2005 09:59 > To: yeoh(a)cs.wisc.edu > Cc: haskell-cafe(a)haskell.org > Subject: Re: [Haskell-cafe] Question on "case x of g" when g > is a function > > Because both bit0 and bit1 are free *local* variables > within the case expression. So, they have nothing > to do with your defined functions bit0 and bit1. > > Best regards, > > Salvador. > > yeoh(a)cs.wisc.edu wrote: > > >Can a kind soul please enlighten me on why f bit0 and f bit1 > >both return 0? > > > > > > > >>bit0 = False > >>bit1 = True > >>f x = case x of > >> bit0 -> 0 > >> bit1 -> 1 > >> ----------------------------------------- ***************************************************************** Confidentiality Note: The information contained in this message, and any attachments, may contain confidential and/or privileged material. It is intended solely for the person(s) or entity to which it is addressed. Any review, retransmission, dissemination, or taking of any action in reliance upon this information by persons or entities other than the intended recipient(s) is prohibited. If you received this in error, please contact the sender and delete the material from any computer. *****************************************************************

1 0

Problem with Hyperlinking Haddock-documentation
by Daniel Fischer 27 Jan '05

27 Jan '05

Maybe somebody can enlighten me. When I run haddock and put the html files e.g. in directory ~/bar/foo, any references to things defined in the Prelude or the libraries are linked to, say ~/bar/foo/Prelude.html#t%3AFractional, which of course does not exist, because the documentation for the Prelude is elsewhere. How can I tell haddock to link such references to where Prelude.html actually is? Please be aware of the fact that I know ///very/// little about html, hyperlinks and such, so be detailed and don't presuppose much. Thanks, Daniel

2 3

List manipulation
by Dmitri Pissarenko 27 Jan '05

27 Jan '05

Hello! I have two lists of Double with equal length and want to create a third one, in which each element is the sum of the corresponding element of the first list and the second list. If list1 is [1, 2, 100] and list2 is [2, 3, 500], then the result of the operation I desire is [3, 5, 600]. I wrote this function <function> add2Img :: [Double] -> [Double] -> [Double] add2Img summand1 summand2 = sum where sum = [ (x+y) | x <- summand1, y <- summand2 ] </function>, but I'm getting [3.0,4.0,501.0,4.0,5.0,502.0,102.0,103.0,600.0] instead of [1, 2, 100]. What is wrong with the function above? TIA Dmitri Pissarenko -- Dmitri Pissarenko Software Engineer http://dapissarenko.com

7 9

QuickCheck properties for Haskell98 Libraries?
by John Meacham 27 Jan '05

27 Jan '05

I was curious if anyone had a collection of QuickCheck rules for testing the Haskell98 Prelude and libraries? It seems like it would be a useful thing to exist as a common resource. In particular, if there were overloaded rules for the various class properties then it would be easier to test whether the instances one writes for standard classes were correct. In fact, it would be nice if the author of every class provided some quickcheck rules for testing the laws instances should obey. John -- John Meacham - ⑆repetae.net⑆john⑈

1 0

What are the MonadPlus laws?
by Jules Bean 26 Jan '05

26 Jan '05

So, anyone? What are the laws that MonadPlus is supposed to satisfy? The obvious ones are that if MonadPlus m then for all types a, (m a) should be a monoid. But, what about the others, because IO does not appear to satisfy a >> mzero == mzero Jules

8 11

RE: [Haskell-cafe] Re: Hugsvs GHC (again)was: Re: Somerandomnewbiequestions
by Simon Marlow 26 Jan '05

26 Jan '05

On 19 January 2005 16:58, Keean Schupke wrote: > Simon Marlow wrote: >> This is what GHC does, if I understand you correctly. The thread >> running select() does so in its own OS thread, while another OS >> thread runs the Haskell code. As long as you use -threaded, that >> is. Oh, and before GHC 6.4 it was done a different way - the >> scheduler used to do the select() between running Haskell threads. >> >> Cheers, >> Simon >> >> > So this means even though the IO calls block, the other Haskell > threads (when run with -threaded) keep running? Yes, unless the IO is to/from disk on a Unix system. Cheers, Simon

8 11

using Map rather than FiniteMap
by S. Alexander Jacobson 26 Jan '05

26 Jan '05

Just did a search after my last post and learned that FiniteMap is bad. Discoverd that Data.Map is the intended replacement. Downloaded it and modified it to work with 6.2. Blazingly fast! Yay. Please ignore the prior message. -Alex- ______________________________________________________________ S. Alexander Jacobson tel:917-770-6565 http://alexjacobson.com

5 7

RE: [Haskell-cafe] using Map rather than FiniteMap
by Simon Marlow 26 Jan '05

26 Jan '05

On 26 January 2005 16:42, S. Alexander Jacobson wrote: > On Wed, 26 Jan 2005, Simon Marlow wrote: >> When using the ordinary 2-generation collector, memory in use will >> tend to be 2-3 times larger than the actual residency, because this >> is a copying collector. > > Ok. Perhaps you want a different name for > this item in the reports? Well, "memory in use" is pretty accurate - it measures the actual amount of memory that the runtime has allocated from the OS. "residency" is a GC term which refers to the amount of live data. I don't think these terms are that confusing, but perhaps my explanation was? >>> On the other hand, I found 10mb, 5mb, and 2.5mb >>> maximum residency, but that is still ~100 bytes >>> per record. > > But 100 bytes per record in a (Map Int Int) seems > awefully high. That 100 bytes includes stack overhead. Remember, when you ran the heap profile which didn't show stack overhead you got a much more reasonable figure. >>> Lastly, I tried "example +RTS -p -K5M -hc" and >>> then looked at the resulting graph (attachment #2) >>> and it shows a total of 1.6Mb heap allocated and >>> if that data is correct, it does correspond >>> roughly to our 20-30 bytes per record estimate. >> >> That profile doesn't include the stack, which sounds reasonable. > > So I guess all the extra memory is actually being > allocated on the stack rather than the heap which > is a sign of a spaceleak. If I solve the > spaceleak I get the 70 bytes per item back. Yup. >> I haven't looked at your code in detail, hopefully someone else can >> shed some light on why you're building up such a large stack in the >> first place. > > The code is just this 6 line program: > > import qualified Map > zipped =zip [1..] [1..100000]::[(Int,Int)] > untup f (x,y) = f x y > produce = foldr (untup Map.insert) Map.empty zipped > fm = length $ Map.keys produce > main = print $ fm > > Can someone tell me what I can do to make this > more efficient? Replace the foldr by a foldl'? import Data.List produce = foldl' (untup (flip Map.insert)) Map.empty zipped (untested) Cheers, Simon

1 0

RE: [Haskell-cafe] using Map rather than FiniteMap
by Simon Marlow 26 Jan '05

26 Jan '05

On 26 January 2005 14:29, S. Alexander Jacobson wrote: > Ah, ok. So I ran the code with 100000 items, > 50000 items, and 25000 items and got total memory > in use of 28Mb, 15Mb, and 8Mb respectively. That > comes to 260-280 bytes per record which is still > an order of magnitude higher than the 20-30 bytes > per record we would expect. When using the ordinary 2-generation collector, memory in use will tend to be 2-3 times larger than the actual residency, because this is a copying collector. > On the other hand, I found 10mb, 5mb, and 2.5mb > maximum residency, but that is still ~100 bytes > per record. Right. > Lastly, I tried "example +RTS -p -K5M -hc" and > then looked at the resulting graph (attachment #2) > and it shows a total of 1.6Mb heap allocated and > if that data is correct, it does correspond > roughly to our 20-30 bytes per record estimate. That profile doesn't include the stack, which sounds reasonable. > Regarding stack, I tried "example +RTS -p -K5M -xt > -hc" and then ran hp2ps and looked at the > resulting graph (attachment #1) SYSTEM appears to > use 4mb of stack at the very beginning (presumably > to create "zipped"), but I don't know why it > would. Then this stack is consumed by the rest of > the program. Stacks never get smaller in GHC, only bigger. If you need 4Mb of stack at one point in your program, you will forever have a 4Mb stack after that (fixing this wouldn't really buy you much; the memory doesn't get traversed or copied by the GC - but one thing you could do is madvise() to tell the OS it doesn't have to page the memory, though). I haven't looked at your code in detail, hopefully someone else can shed some light on why you're building up such a large stack in the first place. Cheers, Simon

2 1

RE: [Haskell-cafe] using Map rather than FiniteMap
by Simon Marlow 26 Jan '05

26 Jan '05

On 25 January 2005 23:27, S. Alexander Jacobson wrote: > Oops. It pays to check your checking code before > making posts like this. > > After actually running the correct test, I am > still getting semi-ridiculous space behavior > (6k/pair)! > > import qualified Map > zipped =zip [1..] [1..100000]::[(Int,Int)] > untup f (x,y) = f x y > produce = foldr (untup Map.insert) Map.empty zipped > fm = length $ Map.keys produce > main = print $ fm > > Has this profile: > > example +RTS -p -K5M -RTS > > total time = 5.10 secs (255 ticks @ 20 ms) > total alloc = 612,534,236 bytes (excludes profiling overheads) > > COST CENTRE MODULE %time %alloc > balance Map 71.8 72.8 > insert Map 12.2 10.8 > size Map 9.0 9.7 > bin Map 2.4 2.2 > rotateR Map 1.6 0.3 > zipped Main 0.8 1.9 > > Notice the 612Mb for storing a mapping from Int > to Int!! Notice that's 612Mb *allocation* to create the finite map and deconstruct it into a list. Not 612Mb of live heap to store the finite map. If you make sure everything is evaluated, and examine the heap profile I'm sure you'll find that the finite map is taking a reasonable amount of space (perhaps 20-30 bytes per node for storing integers, I guess). To get a rough idea of the total live heap without profiling, you can run the program with +RTS -sstderr and look at the memory "in use" figure. Cheers, Simon

2 1