- Haskell-Cafe - Haskell.org

Problem with stack
by Seph Shewell Brockway 24 Feb '21

24 Feb '21

Hi all, A system update (on Arch Linux) seems to have broken my stack installation. It’s started producing linker errors that look like this: dns > /usr/bin/ld.gold: error: /mnt/data/home/seph/.stack/snapshots/x86_64-linux-tinfo6/0ae0d61c27cd61876740871252be93ed57cf041b2170a9014507e77ea977085d/8.10.4/lib/x86_64-linux-ghc-8.10.4/libHScryptonite-0.27-G0nFpwexMw39Vj7zbfAUBq-ghc8.10.4.so:1:1: invalid character dns > collect2: error: ld returned 1 exit status dns > `gcc' failed in phase `Linker'. (Exit code: 1) This seems to happen no matter what snapshot I’m using, although not necessarily with the same packages. Does anyone have any idea what might be going on? Regards, Seph -- Seph Shewell Brockway, BSc MSc (Glas.) Pronouns: she/her

2 3

[ANN] cabal-install-3.4.0.0
by Emily Pillmore 24 Feb '21

24 Feb '21

The Cabal team is excited to announce the release of cabal-install 3.4. This is the third release of the 3.0 release series, bringing a support for the `Cabal-Version: 3.4` `.cabal` file format. This introduces a number of user-facing features including: - Support for GHC 9.0 - Package sublibrary dependency syntax - Allowing omission of the `default-language` declaration - A mixins field allowing the user to specify a sublibrary - [Many other features]( https://cabal.readthedocs.io/en/latest/file-format-changelog.html#cabal-ver… ) In addition, the `cabal-install` frontend itself has gained a number of new features including: - New `active-repositories` `cabal.project` fields allowing users to specify an order and various ways of combining package repositories - Improvements to `cabal init` - A rework of the `cabal v2-sdist` command. `cabal v1-sdist` is removed as of this version - `source-repository-package` directories are no longer local - A new `list-bin` command - `cabal sandbox` is now removed in favor of default nix-style builds - `cabal list` now accepts regular expressions - More rich `index-state` syntax. `v2-freeze` saves `index-state` I'd also like to thank the many contributors who offered patches, tickets, and other help in the preparation of this release. We appreciate all of your help! For a full set of release notes, see https://github.com/haskell/cabal/blob/master/release-notes/cabal-install-3.…. If you have issues, we'd love to hear about there here: https://github.com/haskell/cabal/issues. Happy hacking! Emily

2 2

Cabal 3.2.0.0, run okay with profiling enabled, but installed exe NO profiling enabled
by YueCompl 24 Feb '21

24 Feb '21

Dear Cafe, I'm not sure it is a bug or just designed behavior? But I can't install profiling-enabled exe with cabal 3.2.0.0, while direct running works correctly. $ cabal --version cabal-install version 3.2.0.0 compiled using version 3.2.0.0 of the Cabal library $ cabal run --enable-profiling --enable-executable-profiling exe:els -- +RTS --info Up to date [("GHC RTS", "YES") ,("GHC version", "8.8.4") ,("RTS way", "rts_thr_p") ,("Build platform", "x86_64-apple-darwin") ,("Build architecture", "x86_64") ,("Build OS", "darwin") ,("Build vendor", "apple") ,("Host platform", "x86_64-apple-darwin") ,("Host architecture", "x86_64") ,("Host OS", "darwin") ,("Host vendor", "apple") ,("Target platform", "x86_64-apple-darwin") ,("Target architecture", "x86_64") ,("Target OS", "darwin") ,("Target vendor", "apple") ,("Word size", "64") ,("Compiler unregisterised", "NO") ,("Tables next to code", "YES") ,("Flag -with-rtsopts", "-maxN31 -qg") ] $ cabal install --enable-profiling --overwrite-policy=always --enable-executable-profiling exe:els && els +RTS --info Wrote tarball sdist to ... Resolving dependencies... Up to date Copying 'els' [("GHC RTS", "YES") ,("GHC version", "8.8.4") ,("RTS way", "rts_thr") ,("Build platform", "x86_64-apple-darwin") ,("Build architecture", "x86_64") ,("Build OS", "darwin") ,("Build vendor", "apple") ,("Host platform", "x86_64-apple-darwin") ,("Host architecture", "x86_64") ,("Host OS", "darwin") ,("Host vendor", "apple") ,("Target platform", "x86_64-apple-darwin") ,("Target architecture", "x86_64") ,("Target OS", "darwin") ,("Target vendor", "apple") ,("Word size", "64") ,("Compiler unregisterised", "NO") ,("Tables next to code", "YES") ,("Flag -with-rtsopts", "-maxN31 -qg") ] $ cabal install --enable-profiling --overwrite-policy=always --enable-executable-profiling exe:els && els +RTS -p Wrote tarball sdist to ... Resolving dependencies... Up to date Copying 'els' els: the flag -p requires the program to be built with -prof els: els: Usage: <prog> <args> [+RTS <rtsopts> | -RTS <args>] ... --RTS <args> els: ... Is there other options to get a profiling enabled exe installed? Thanks with regards, Compl

2 2

[ANN] zlib-0.6.2.3
by Emily Pillmore 22 Feb '21

22 Feb '21

Hello Everyone, I'm pleased to announce zlib-0.6.2.3, which adds support for the bytestring-0.11 work, and unblocks a decent number of packages to the next bytestring versions. I would also like to announce that two new maintainers have been added to the zlib team: Andrew Lelechenko (bodigrim), and myself (topos). Cheers, Emily

1 0

Munich Virtual Haskell Meeting, 2021-02-23 @ 19:30
by Leah Neukirchen 21 Feb '21

21 Feb '21

Dear all, Next week, our monthly Munich Haskell Meeting will take place again on Tuesday, Feburary 23 on Google Meet at 19:30 CET. **Due to meetup limitations in Bavaria, this meeting will take place online!** For details see here: https://muenchen.haskell.bayern/dates.html A Google Meet link to join the room is provided on the page. Everybody is welcome, especially the Haskellers from Bavaria that do not usually come to our Munich meetings due to travel distance! cu, -- Leah Neukirchen <leah(a)vuxu.org> https://leahneukirchen.org/

1 0

dyre-0.9.0 release candidate and call for testing
by Fraser Tweedale 21 Feb '21

21 Feb '21

Dyre is a tool for Haskell program configuration by (re)compilation. After years of project dormancy, I am pleased to announce a Dyre 0.9 release candidate[1]. If you are interested and able to test this candidate release, you can report results at [2]. For more details including changelog, see the accompanying blog post[3]. [1] https://hackage.haskell.org/package/dyre-0.9.0/candidate [2] https://github.com/willdonnelly/dyre/issues/39 [3] https://frasertweedale.github.io/blog-fp/posts/2021-02-21-dyre-0.9-rc.html Thanks, Fraser

1 0

More Efficient Recursion
by A. Mc. 21 Feb '21

21 Feb '21

Hello, I'm just starting off, but I'm wondering if anyone has any suggestions. I've made a recursive function that's goal is to handle a base raised to a very large power. modLoop :: Integer -> Integer -> Integer -> Integer -> Integer -> Integer modLoop iterator exponent result base modulus | (iterator == exponent) = result | (iterator < exponent) = modLoop (iterator + 1) (exponent) (mod (result * base) modulus) base modulus | (iterator > exponent) = (-1) However, it flat breaks or takes a long time when using large numbers. I need to get better at using strict functions, however, I compiled with ghc -o (as it tends to implement the strict version - ?) and I still have the same problem. What's a better way? Thanks in advance and thank you for your time.

3 2

help with existential type signature?
by Ken T Takusagawa 20 Feb '21

20 Feb '21

{- I am experimenting with type-level naturals and with moving values known at runtime to type space. Is there a type signature that I can add to f1 below so that this will compile (maybe also requiring some additional LANGUAGE pragmas)? If I were to change the return value of f1 from "x" to "show x", then the program does compile, suggesting that there's nothing inherently wrong with "x". But I'd prefer the option to compose the output of f1 with other computations separately, e.g., "show . f1" . Thanks, Ken -} {-# LANGUAGE KindSignatures, ScopedTypeVariables, DataKinds, Rank2Types #-} import qualified GHC.TypeLits as TypeLits; import Data.Proxy(Proxy); main = undefined; data Foo (fn :: TypeLits.Nat) = Mkfoo deriving (Show); --f1 :: TypeLits.SomeNat -> ? ; f1 mynat = case mynat of { TypeLits.SomeNat (_ :: Proxy n) -> let { x :: Foo n; x = Mkfoo; } in x }; {- typelits.hs:27:8: error: * Couldn't match expected type `p' with actual type `Foo n' because type variable `n' would escape its scope This (rigid, skolem) type variable is bound by a pattern with constructor: TypeLits.SomeNat :: forall (n :: TypeLits.Nat). TypeLits.KnownNat n => Proxy n -> TypeLits.SomeNat, in a case alternative at smalltypelits2.hs:24:3-33 * In the expression: x In the expression: let x :: Foo n x = Mkfoo in x In a case alternative: TypeLits.SomeNat (_ :: Proxy n) -> let x :: Foo n x = Mkfoo in x * Relevant bindings include x :: Foo n (bound at smalltypelits2.hs:26:5) f1 :: TypeLits.SomeNat -> p (bound at smalltypelits2.hs:23:1) | 27 | } in x | ^ -}

2 1

Grapheme length?
by amindfv＠mailbox.org 20 Feb '21

20 Feb '21

Does there exist a Haskell library or function for getting grapheme lengths of String/Text values? e.g. right now, Prelude.length ("ä" :: String) == 2 Data.Text.length ("ä" :: Text) == 2 But I'd like, either for String or Text, to get 1 instead (there are 2 code points but only 1 grapheme). Thanks, Tom

2 6

Sizes in memory?
by Dušan Kolář 20 Feb '21

20 Feb '21

Dear café, I don't know, whether this is the right place to ask, but I'll try :-) If I would like to know size of the list in the memory (after complete evaluation, e.g. deepseq). The type is [( Int, [([Int],[Int])] )]. So we have K tuples of the type (Int,[([Int],[Int])]) Each tuple has N tuples of the type ([Int],[Int]) and each of these lists M elements long (yes, both a are of the same length). Top level list needs (on 64bit CPU) 8 bytes for pointer to elements, 8 bytes to point to another element, some more "expenses", probably yes, tag, anything else? So we have K*(16+XXX), where XXX is for size of (Int,[([Int],[Int])]). Tuple needs 8 bytes for each pointer, so XXX = 16 + YYY, where YYY is the size of the [([Int],[Int])] list. It needs 8 + 8 bytes on the list, we have N such elements, thus YYY = N * (16 + ZZZ), where ZZZ is size of the ([Int],[Int]). Again tuple, so 16 bytes + 2*(M * (16 + 8)) -- 16 bytes for 2 tuple elements -- 2 for two lists -- every list 16 bytes for list itself -- 8 bytes for Int So ZZZ = 16 + 2*(M * (16 + 8)), thus YYY = N * (16 + 16 + 2*(M * (16 + 8))), thus XXX = 16 + N * (16 + 16 + 2*(M * (16 + 8))), thus size = K * (16 + 16 + N * (16 + 16 + 2*(M * (16 + 8)))) = K* (32 + N* (32 + M*48) ) Am I right? Roughly right? Or totally wrong? E.g. for K = 25, N = 10000, M = 7 we get cca 88MiB ? Why am I asking? In reality the memory consumption is much higher, so I must be missing some extras, e.g. each memory chunk extra 8 bytes for GC, extra for tag, extra to point to VMT? Regards, Dušan

2 1