- Haskell-Cafe - Haskell.org

unusual behavior from cabal
by Michael Litchard 18 Jul '10

18 Jul '10

Not sure what the correct list is for this observation. I was trying to install gitit, and here is what happened. michael@michael:~/haskell/blog-example$ cabal install gitit Resolving dependencies... cabal: dependencies conflict: happstack-server-0.5.1 requires time ==1.1.4 however time-1.1.4 was excluded because happstack-server-0.5.1 requires time ==1.2.0.3 huh? Cabal seems to think happstack-server-0.5.1 has two conflicting requirements. How do I suss this out?

3 2

ANN: hledger 0.11
by Simon Michael 17 Jul '10

17 Jul '10

hledger 0.11 is released! Thanks to all testers and to Michael Snoyman for much help keeping up with Yesod. Best, -Simon home: http://hledger.org Release notes: 2010/07/17 hledger 0.11 ........................ * split --help, adding --help-options and --help-all/-H, and make it the default command * use "journal" instead of "ledger file"; default suffix is .journal, default file is ~/.journal * auto-create missing journal files rather than giving an error * new format-detecting file reader (mixed journal transactions and timelog entries are no longer supported) * work around for first real-world rounding issue (test zero to 8 decimal places instead of 10) * when reporting a balancing error, convert the error amount to cost * parsing: support double-quoted commodity symbols, containing anything but a newline or double quote * parsing: allow minus sign before commodity symbol as well as after (also fixes a convert bug) * parsing: fix wrong parse error locations within postings * parsing: don't let trailing whitespace in a timelog description mess up account names * add: allow blank descriptions * balance: --flat provides a simple non-hierarchical format * balance: --drop removes leading account name components from a -- flat report * print, register, balance: fix layout issues with mixed-commodity amounts * print: display non-simple commodity names with double-quotes * stats: layout tweaks, add payee/description count * stats: don't break on an empty file * stats: -p/--period support; a reporting interval generates multiple reports * test: drop verbose test runner and testpack dependency * web: a new web ui based on yesod, requires ghc 6.12; old ghc 6.10- compatible version remains as -fweb610 * web: allow wiki-like journal editing * web: warn and keep running if reloading the journal gives an error * web: --port and --base-url options set the webserver's tcp port and base url * web: slightly better browser opening on microsoft windows, should find a standard firefox install now * web: in a web-enabled build on microsoft windows, run the web ui by default Stats: 55 days and 136 commits since last release. Now at 5552 lines of code with 132 tests and 54% unit test coverage.

1 0

Cabal lib (Setup.hs), --package-db, --enable-shared, haskell platform
by Albert Y.C.Lai 17 Jul '10

17 Jul '10

I accidentally found a rarely encountered omission in Cabal (the lib, because via Setup.hs) by building Haskell Platform as shared libs from source. It is rare because you have to use both --enabled-shared and --package-db=blah together to run into it. --package-db=blah is already rare enough (probably only Haskell Platform's build script uses it); --enable-shared is my addition to make it even rarer. (Also I have only tried this on linux x86 32-bit.) Following the essence of scripts/build.sh plus my own --enable-shared: Create an empty lookaside package database with: echo '[]' > /tmp/lookaside Install a package X that depends only on base (say), ask for shared libs, register in /tmp/lookaside only: ./Setup configure --package-db=/tmp/lookaside \ --ghc-pkg-option=--package-conf=/tmp/lookaside \ --enable-shared ./Setup build ./Setup register --inplace Install another package Y that depends on base and X only (say), ask for shared libs. There is a problem now related to X known to /tmp/lookaside only, but not known to the global or the user database: ./Setup configure --package-db=/tmp/lookaside \ --ghc-pkg-option=--package-conf=/tmp/lookaside \ --enable-shared ./Setup build -v2 Now it bombs. As the verbose build messages show, there is no problem creating the static *.a version; it bombs when creating the shared *.so version, and it is because it can't find X, and it is because -package-conf=/tmp/lookaside is not given to ghc at the *.so stage. (Though, the *.dyn_o stage works correctly.) My current fix is: ./Setup configure --package-db=/tmp/lookaside \ --ghc-pkg-option=--package-conf=/tmp/lookaside \ --ghc-option=-package-conf=/tmp/lookaside \ --enable-shared I surely hope we no longer need --ghc-pkg-option and --ghc-option in a week! :)

1 1

Re: [Haskell-cafe] is this a bug ?
by gate03＠landcroft.co.uk 17 Jul '10

17 Jul '10

On Sat 17/07/10 04:17 , Alexander Solla ajs(a)2piix.com sent: > Why are you performing unsafe IO actions? They don't play nice > with laziness. OK, fair cop, but without the unsafe IO action, it still misbehaves. http://hpaste.org/fastcgi/hpaste.fcgi/view?id=27650 Michael.

3 2

Re: in-equality type constraint?
by oleg＠okmij.org 17 Jul '10

17 Jul '10

Ryan Ingram wrote: > But it doesn't generalize; you need to create a witness of inequality > for every pair of types you care about. One can do better, avoiding the quadratic explosion. One merely needs to establish a map from a type to a suitable, comparable representation -- for example, to a type level list of type level numerals. Comparing types for equality, inequality and even order is a simple matter of comparing their representations. The fact that types become totally ordered lets us even implement type-level maps: Data.Map on types. (We may need a Type.* module hierarchy.) That idea was described in the HList paper http://homepages.cwi.nl/~ralf/HList/paper.pdf Section 9. The code is still available, http://code.haskell.org/HList/examples/TTypeable.hs see also TypeEqExplosive.hs, TypeEqTTypeable.hs. > But given the use of UndecidableInstances and OverlappingInstances, I > was hoping that type families could come a little cleaner. Normally if the use of functional dependencies requires UndecidableInstances, type families would ask for them too. As to OverlappingInstances -- that is the key to generic inequality, isn't it? Given the set of pairs of types, the set of the elements representing non-equal types is the complement of the set of elements representing equal types (by equal I mean identical). Overlapping instances is the way to express set complementation. The most general instance is chosen when none of the more specific apply. The set of types chosen by that general instance is the complement for the set of types chosen by the specific instances. > Does "TypeEq a c HFalse" imply proof of inequality, or unprovability > of equality? We are all constructivists here... If 'a' and 'c' are type variables, (TypeEq a c HFalse) is the constraint -- proof obligation if you will -- making sure the variables will never be instantiated to identical types. Strictly speaking, the constraint is discharged if 'a' and 'c' are two ground, and syntactically distinct (non-identical) types. In reality, I think GHC is able to discharge the constraint if 'a' and 'c' are grounded ``sufficiently enough'' for the difference to become apparent (e.g., the head constructors are different).

2 1

ghc api printing of types
by Phyx 17 Jul '10

17 Jul '10

I was wondering how given a Type I can get a pretty printed type out of it. I'm currently using showSDocUnqual . pprType . snd . tidyOpenType emptyTidyEnv But this has the problem that predicates don't get printed, anyone know how GHCi does this? Thanks, Phyx

2 3

Plotting Vectors with GnuPlot Wrapper
by Maryam Moghadas 17 Jul '10

17 Jul '10

Hi When I use Vectors as a PlotStyle in Graphics.Gnuplot.Simple, the output curve.gp and curve0.csv is not generated correctly. For example when I write in ghci: plotPathStyle [] (PlotStyle Vectors (DefaultStyle 1)) [(1,1),(2,7)] the output files are: curve.gp: plot "curve0.csv" using 1:2 with vectors linestyle 1 curve0.csv: 1, 1 2, 7 But they must be: curve.gp: plot "curve0.csv" using 1:2:3:4 with vectors linestyle 1 curve0.csv: 1, 1, 2, 7 Is there any option that should I use to correct this behavior? Regards Mary

2 1

Hot-Swap with Haskell
by Andy Stewart 17 Jul '10

17 Jul '10

Hi all, I'm research to build a hot-swap Haskell program to developing itself in Runtime, like Emacs. Essentially, Yi/Xmonad/dyre solution is "replace currently executing" technology: re-compile new code with new binary entry when re-compile success $ do save state before re-launch new entry replace current entry with new binary entry (executeFile) store state after re-launch new entry There are some problems with re-compile solution: 1) You can't save *all* state with some FFI code, such as gtk2hs, you can't save state of GTK+ widget. You will lost some state after re-launch new entry. 2) Sometimes re-execute is un-acceptable, example, you running some command in temrinal before you re-compile, you need re-execute command to restore state after re-launch, in this situation re-execute command is un-acceptable. I wonder have a better way that hot-swapping new code without re-compile/reboot. Thanks, -- Andy

9 13

in-equality type constraint?
by Paul L 17 Jul '10

17 Jul '10

Does anybody know why the type families only supports equality test like a ~ b, but not its negation? -- Regards, Paul Liu Yale Haskell Group http://www.haskell.org/yale

7 8

ANNOUNCE: feldspar 0.3
by Emil Axelsson 17 Jul '10

17 Jul '10

On behalf of the Feldspar team, I'm happy to announce a new release of the embedded language Feldspar and its C code generator: http://hackage.haskell.org/package/feldspar-language http://hackage.haskell.org/package/feldspar-compiler The main changes in 0.3 are: * Signed/unsigned integers of different bit widths (8, 16 and 32). * Fixed-point numbers. * Slight change of the vector library interface: All intermediate vectors are guaranteed to be removed. * A new Stream data type. * Support for user-defined types. * Tracing functions. * Limited support for code generations for the TMS320C64x chip family. * Bugfixes. For more information, see: http://feldspar.inf.elte.hu/feldspar/ / Emil

1 0