November 2015 - Haskell-Cafe

checking whether the list is active
by Miroslav Kocic 12 Nov '15

12 Nov '15

3 2

types calculation
by Dmitry Olshansky 11 Nov '15

11 Nov '15

Hello! It seems that types without parameters are not reduced in ghc unlike CAFs. I.e. if we have type T1 a b = ... type T2 = T1 Int Int than T2 will be calculated on each utilization. Is my statement correct? If so, why is it? With complicated type-calculation compile time is growing too fast. Best regards, Dmitry

2 4

Stack: Copying possible?
by Rustom Mody 11 Nov '15

11 Nov '15

Trying to switch over to stack I copied over ~/.stack to another Linux box And it gives me sanity check failed Machines are same -- both 64 bit ubuntu wily -- user-names are different and that shows in the error message. I'd like to avoid almost a Gb of repeated downloads if possible. But if thats not kosher that fine... Just asking :-)

3 4

Question, re: illegal instance error.
by David Banas 11 Nov '15

11 Nov '15

Hi all, I don’t understand why this code: 50: instance (IsNat n, RealFloat a) => FFT (RTree n) a where is generating this error: fft_test.hs:50:36: Illegal instance declaration for ‘FFT (RTree n) a’ (All instance types must be of the form (T a1 ... an) where a1 ... an are *distinct type variables*, and each type variable appears at most once in the instance head. Use FlexibleInstances if you want to disable this.) In the instance declaration for ‘FFT (RTree n) a’ and was hoping someone could help me get it. Thanks! -db

3 2

[ANN] SBV 5.4 is out
by Levent Erkok 10 Nov '15

10 Nov '15

I'm happy to announce a new release (v5.4) of SBV: SMT Based Verification. See: http://hackage.haskell.org/package/sbv New in this release is the 'sAssert' function, which allows boolean-assertions to be sprinkled in symbolic code, which can then be statically discharged (or shown violating assignments for) using the 'safe' function. See: http://hackage.haskell.org/package/sbv-5.4/docs/Data-SBV.html#g:39 A slightly larger example checks for lack-of-division-by-0 in a simple program: http://hackage.haskell.org/package/sbv-5.4/docs/Data-SBV-Examples-Misc-NoDi… Further documentation on SBV is available at: http://leventerkok.github.io/sbv/ Calls to sAssert calls are also preserved in the SBV->C compilation path. This feature is rather experimental and , and I'd love to hear if you give it a try and see any gotchas.. Happy proving. -Levent. PS. Thanks to Brett Letner for suggesting this functionality.

1 0

Anybody using the "top-down" cabal-install solver?
by Bardur Arantsson 09 Nov '15

09 Nov '15

(Sorry if you get this twice, forgot to cross-post to -cafe initially.) Hi all, Just to get input from as many people as possible: I was pondering a plan for modularizing[2] the Cabal solver[1] and wanted to reach as many people as possible with my question: Is anybody is still using the top-down solver? Please respond to this list if you are, especially if you're doing so because there's no other way to get the modular solver to do what you want. (Obviously, I'm not promising to fix any problems you may have with the modular solver, but it would be valuable information since it could affect the decision of whether we have to keep the top-down solver around.) Regards, -- [1] https://github.com/haskell/cabal/pull/2768#issuecomment-154917165 [2] "Splitting it out into its own library" is perhaps a more accurate description, but somewhat unwieldly :).

1 0

Where io-streams won't fit, comparing to conduit or pipes
by Daisuke Fujimura 09 Nov '15

09 Nov '15

Hello cafe, Now I'm preparing a presentatio for meetup for stream processing in functional programming(in Tokyo), and my part is something like "why stream processing library". While preparing material, I encountered the question that why io-streams is so simple comparing others, and the reason and downsides behind its simplicity. I'm guessing it's due to its resource handling strategy, but not so confident about it because I couldn't came up with a concrete example yet. Do you know any good example where io-streams won't fit? Let me know if you have. Thanks!

4 3

How to get GHC to produce ADC instructions for long addition
by Clinton Mead 09 Nov '15

09 Nov '15

Here is some code that adds two 192 bit numbers, represented as three 64bit machine words (well, on my machine anyway), and returns the result and any carry: {-# LANGUAGE MagicHash #-} {-# LANGUAGE UnboxedTuples #-} import GHC.Prim( plusWord2#, Word#, or#) longAdd :: (# Word#, Word#, Word# #) -> (# Word#, Word#, Word# #) -> (# Word#, (# Word#, Word#, Word# #) #) longAdd (# xl, xm, xh #) (# yl, ym, yh #) = let plusWord2WithCarry x y c = let (# c1, r1 #) = plusWord2# x y (# c2, r2 #) = plusWord2# r1 c in (# plusWord# c1 c2, r2 #) (# cl, rl #) = plusWord2# xl yl (# cm, rm #) = plusWord2WithCarry xm ym cl (# ch, rh #) = plusWord2WithCarry xh yh cm in (# ch, (# rl, rm, rh #) #) (My code covers words other than size 3 btw) I'd like this to compile into something like: add x1 y1 adc x2 y2 adc x3 y3 Unfortunately, I think my problem is with the "plusWord2WithCarry". As there's no primitive operation in Haskell which adds two words and a carry. What seems to happen when I look at the generated assembly is the following: xor some_reg_1 some_reg_1 add x1 y1 adc $0 some_reg_1 xor some_reg_2 some_reg_2 adc x2 y2 adc $0 some_reg_2 xor some_reg_3 some_reg_3 adc some_reg_1 y2 adc $0 some_reg_3 add some_reg_2 some_reg_3 ... and so on Basically, instead of just using the carry flag in add of the next two higher words, it instead saves it to a register, adds the next higher order words without the carry, clears a register and then adds the carry to this register, adds the first mentioned carry, against clears a register and saves that carry, combines the two resulting carries and passes it to the next higher order addition. I know that sounds complex, and the code is a bit of a mess. Bizarrely, when I send it through the LLVM backend it ends up worse, generating a bunch of 32 bit shifts for reasons I can't understand. I was hoping the LLVM backend would be able to produce the "adc" instructions. Is there anything I could do to coax it into it. I know writing it in C (or inline assembly in C) is an option, but after you add the code to call and return such a small amount of work it seems hardly worth it. I'd like to keep it in GHC so it can be inlined where appropriate. Any ideas on how to entice GHC (either natively or via LLVM) to produce better code in this case?

1 0

More death to partial functions: Should -fwarn-incomplete-uni-patterns be enabled with -Wall?
by Niklas Hambüchen 08 Nov '15

08 Nov '15

Hello everyone, to my dismay, a -Wall compiled program I deployed today crashed with Exception: Non-exhaustive patterns in lambda and I learned from http://dev.stephendiehl.com/hask/ that A more subtle case is when implicitly pattern matching with a single "uni-pattern" in a lambda expression. The following will fail when given a Nothing. boom = \(Just a) -> something GHC can warn about these cases with the -fwarn-incomplete-uni-patterns flag. And in fact: ghci -Wall > map (\Nothing -> "ok") [Just 'a'] ["*** Exception: <interactive>:2:6-21: Non-exhaustive patterns in lambda > :set -fwarn-incomplete-uni-patterns > map (\Nothing -> "ok") [Just 'a'] <interactive>:4:6: Warning: Pattern match(es) are non-exhaustive It really surprised me that -fwarn-incomplete-uni-patterns is not included in -Wall; I've felt really safe with my -Wall so far, especially about totality and similar things that will almost certainly lead to crashes. In an older mail from 2010 (https://mail.haskell.org/pipermail/glasgow-haskell-users/2010-September/019…) I saw Simon mention that it wasn't planned to warn about inexhaustive patterns like this. I feel like there has been a strong push towards more totality in the last years, and I would like to ask if enabling -fwarn-incomplete-uni-patterns in -Wall may be reconsidered for a coming next GHC version, or if there are still opponents to that. Niklas

9 10

Pattern matching with OverloadedLists
by Vlatko Basic 06 Nov '15

06 Nov '15

Hello Cafe, wiki for OverloadedLists says that g [x,y,z] = ... is treated as g (toList-> [x,y,z]) = Shouldn't this work? Both 'f's should be treated the same. f :: Text -> Int f (toList -> ('9':_)) = 1 -- OK f ('1':_) = 2 -- Couldn't match expected type ‘Text’ with actual type ‘[Char]’ (OverloadedStrings is also on) Am I missing something? br, vlatko

2 2