- Haskell-Cafe - Haskell.org

Cyclic Type-synonyms
by Christophe Poucet 05 Jul '06

05 Jul '06

Hello, I use the Indirect Composite pattern a lot, and this means that typically, especially with recursive types (such as an AST), you end up with a lot of data-constructors. I understand that it is not possible to have pure cyclic types (because haskell requires iso-recursive and not equi-recursive types). However I wonder why certain cases can not be allowed. For instance, assume the following somewhat simplified AST: data Exp e = ELet String e e -- let x = a in b | ECond e e e -- If then else construct Now if I wanted to decorate this structure, I would use something like: data AnnotExp a = AE { unAE :: Exp (AnnotExp a), note :: a } However, and this example might be slightly too trivial to motivate it, however it does exemplify what can be done, one might not want to annotate at all and just have a recursive data AST. At this point, one either uses AnnotExp () or creates a new data type. It would be nice if instead one could use cyclic type-declarations, with the knowledge that we're going through a data-constructor and hence the type iso-recursive and not equi-recursive: type AST = Exp AST -- This is iso-recursive as it goes through the data constructors ELet and ECond Sadly this is not possible with the current GHC (I'm using 6..4.2). Is this possible with 6.6? Is there any reason why this is not possible. Feedback regarding this is welcome. A more motivating example can be seen below. Cheers, Christophe ----------------------------- data Exp var ident func exp stm = ELit Lit -- ^ Literals | EVar ident -- ^ Identifiers | EData Const [exp] -- ^ Data Constructors | ECall func -- ^ Function Calls | ELet var stm stm -- ^ Scoping | ECond exp stm stm -- ^ Conditional | ELoop exp stm -- ^ Looping deriving (Eq, Show) data AExp exp annot = AExp { unAExp :: exp, aexpNote :: annot, aexpLoc :: Location } type UnCorExp var annot = Exp var -- ^ Let-binders var -- ^ Named identifiers (Ident, [AExp UnCorExp var annot]) -- ^ Function calls (AExp UnCorExp var annot) -- ^ Expressions (AExp UnCorExp var annot) -- ^ Statements -- Flattened AST: function parameters can only be variables, similar for while- and if- conditions type UnSimExp var annot Exp var -- ^ Let-binders var -- ^ Named identifiers (Ident, [var]) -- ^ Function calls (var) -- ^ Expressions (AExp UnSimExp var annot) -- ^ Statements

1 1

Clever generic ByteString hack?
by John Goerzen 05 Jul '06

05 Jul '06

Hi, In MissingH, I have a bunch of little functions that operate on lists. Some, like uniq (which eliminates duplicate elements in a list), operate on (Eq a => [a]) lists. Others, like strip (which eliminates whitespace at the start and end), operate on Strings only. Most functions of both types would be useful on ByteStrings and lazy ByteStrings. Most of these functions are written in terms of Data.List functions or list primitives that have equivolents in Data.ByteString. So, my question is: is there a clever hack available to me so that I could have 1 version of each function, and have it work on all three different types of input? I'd rather avoid having 3 versions, that are exactly the same except for imports. Thanks, -- John

4 6

Re: [haskell] ANNOUNCE: HNOP 0.1
by mvanier 05 Jul '06

05 Jul '06

[Moving to haskell-cafe] OK, after going through this several times, here's what I've found: 1) The Debian linux build of haskell produces large executables. 2) The generic x86 binary distribution produces smallish executables (383765 bytes; still kind of large; stripped it's 191584 bytes). 3) I removed my entire source distribution, and recompiled from scratch with this build.mk file: SRC_HC_OPTS = -H64m -O2 GhcLibHcOpts = -O2 -fgenerics SplitObjs = YES I did an "autoreconf; configure; make; make install" and then recompiled hnop from scratch. I got large executables (2838528 bytes unstripped, 1540072 stripped). I'm at a loss here. Somehow, the SplitObjs option doesn't seem to be doing the job. Any suggestions would be appreciated. Mike Donald Bruce Stewart wrote: > Hmm. Did you build from scratch? Perhaps the libraries didn't get > rebuilt with split objs ? > > > mvanier: >> I've tried doing a manual build with SplitObjs set to YES, but without >> success; either the compiler was broken or it would still generate large >> executables. So as a sanity check, I installed the debian unstable build. >> But that one also gives me a huge executable: >> >> % ls -l >> total 2720 >> -rw-r--r-- 1 mvanier mvanier 818 Jun 29 20:48 LICENSE >> -rw-r--r-- 1 mvanier mvanier 308 Jun 30 04:59 Main.hi >> -rw-r--r-- 1 mvanier mvanier 80 Jun 29 20:48 Main.hs >> -rw-r--r-- 1 mvanier mvanier 1928 Jun 30 04:59 Main.o >> -rw-r--r-- 1 mvanier mvanier 43 Jun 29 20:48 Makefile >> -rwxr-xr-x 1 mvanier mvanier 2756151 Jun 30 04:59 hnop >> -rw-r--r-- 1 mvanier mvanier 807 Jun 29 23:56 hnop.tar.gz >> >> Am I the only person in the world with this problem? >> >> Also, for some inexplicable reason, debian puts ghci into /usr/X11R6/bin, >> though ghc is in the more sensible /usr/bin/ghc. >> >> TIA, >> >> Mike >> >> Donald Bruce Stewart wrote: >>> mvanier: >>>> Nope, just ghc-6.4.2 that I compiled myself on a Dell x86 laptop (Pentium >>>> M). I didn't use any strange configure options. Is there a special >>>> configure flag for split objects? If not, how is this determined? >>> Ah ha! You need to add -split-objs to the build.mk used by ghc to build >>> its libraries, as described here: >>> http://cvs.haskell.org/cgi-bin/cvsweb.cgi/~checkout~/fptools/ghc/HACKING?co… >>> >>> You want: >>> SplitObjs = YES >>> >>> in mk/build.mk >>> >>> It's not enabled by default, since it doesn't work on every platform. >>> The ghc that is distributed in the debian package system does have split >>> objs on, though. >>> >>> Cheers, >>> Don

3 4

Where is Data.Atom ?
by Brian Hulley 04 Jul '06

04 Jul '06

Hi - I seem to remember from some discussion long ago that there is a Data.Atom module somewhere (hopefully with a BSD licence), but a search on Hoogle turned up no results so maybe I'm mistaken. What I'm looking for is an Atom type and the following operations: fromString :: String -> Atom toString :: Atom -> String instance Eq Atom instance Ord Atom -- this is where things get difficult! It is particularly important that the ordering comparison is extremely fast, preferably independent of the length of the original String, and that it preserves the lexicographic ordering between the original Strings. I can see that an "unsafe" global ref to a Trie of Char with Unique as the "value" of a node would allow me to implement fromString, toString, and instance Eq Atom, but I've got no idea how to implement instance Ord Atom so that the order is independent of the order in which Atoms are created and exactly the same as the lexicographic ordering of the String without being O(n) where n is the min of the lengths of the Atoms being compared. I'm also hoping that Atoms which are no longer in use would manage to magically vanish by themselves. In a Trie implementation, this would mean maintaining an invariant that leaf nodes are always held by a weak link and internal nodes by a strong link. Any ideas (or pointers to a nice downloadable module that already does all this :-) )? Thanks, Brian. -- Logic empowers us and Love gives us purpose. Yet still phantoms restless for eras long past, congealed in the present in unthought forms, strive mightily unseen to destroy us. http://www.metamilk.com

8 14

HDF5
by Joel Reymont 04 Jul '06

04 Jul '06

Does anyone have bindings for HDF5 [1]? [1] http://hdf.ncsa.uiuc.edu/whatishdf5.html Thanks, Joel -- http://wagerlabs.com/

2 3

EclipseFP (Haskell IDE) 0.10.0 released
by Thiago Arrais 04 Jul '06

04 Jul '06

EclipseFP 0.10 has been released since last Friday. It is an open-source development environment for Haskell code. EclipseFP integrates GHC with an Haskell-aware code editor and also supports quick file browsing through an outline view, automatic building/compiling and quick one-button code execution. For this new release we have been working on code assistance support, something that has been heavily requested, and a Cabal editor. Downloads and more information are available on the project home page http://eclipsefp.sourceforge.net/ This release requires Eclipse 3.2, as opposed to Eclipse 3.1 that was required on the previous release. You may need to upgrade your Eclipse platform if you plan to use version 0.10. I have taken the liberty to create a wiki page for EclipseFP. You can post your comments and feature requests there, if you'd like. It is located at http://www.haskell.org/haskellwiki/EclipseFP Cheers, Thiago Arrais

7 12

hquickfile first prototype ..
by Marc Weber 04 Jul '06

04 Jul '06

I want your feedback.. The purpose of this program is to get the files/ directories you know faster. Put this into your .bashrc: Of course you have to adjust the executable filename. ============= .bashrc ========================================================== alias F='/usr/bin/hquickfile' Q(){ cd `/usr/bin/hquickfile "$@"` } ============= end .bashrc ====================================================== The source can be found here: http://www.inversoft.de/hquickfile/hquickfile-0.0.tar.gz You need ghc, cabal, parsec and the FilePath library of Neil (Thanks for assisting me!) http://www-users.cs.york.ac.uk/~ndm/projects/libraries.php Things to try: Q /t (equal to cd /tmp) Q /usl (equal to cd /usr/src/linux) .. you should get the idea by reading the help output: ============ hquickfile -h output ============================================= Usage: hquickfile Get the file you want quickly. hquickfile is not a new shell yet but a beginning.. ;) usage: you specify file patterns such as (explanation in bash glob syntax) @vl which will match v.*/l.* (eg var/log on liunux) :abc:def will match .*a.*b.*c.*/d.*e.*f.* =abc will match abc exactly prefix with / to start from root, not current directory windows users have to use /c for c:, /d for d: .. (win not yet tested) to take the first x directories as relative path to start from prepend x to remove the last x directories prepend x- eg cd is /home/marc/hq/ pattern 1-=foo will result in /home/marc/foo pattern 1=foo will result in /home/foo (match file 1 with =1 or @1 append '+' to add not yet existing parts eg you can do mkdir -p `thisapp /@t+a/b/c` to create the /tmp/a/b/c directory append / or : to match directories or files only you can use {blah,foo,...}/ab as you can do in bash which expands to two arguments blah/ab and foo/ab {a,b}{c,d} will expand to ac ad bc cd wheras {a,b}foo[c,d] will expand to afooc bfood {a,b}{c,d}[1,2,3.4] will expand to ac1 ad2 bc3 cd4 you can add glob patterns: ,a/*.txt will ask you wether you want all text files from abc/*.txt or abd/*.txt Because every key you have to type is important you can omit the leading @ so /ea is equal to /@ea would it be useful if small letters would also match capital letters but not the other way round? The Eclipse Plugin GotoFile (written by Max Uermann) does it this way written by Marc Weber Credits to any other haskeller whose tools I've been using (Neil, Pesco, ..) ! especially those on #haskell on freenode who have helped me a lot by listening -h, --help print this help ========== end hquickfile -h ================================================ I think there are still some bugs.. Marc Weber

1 0

JRegex on "large" input sizes
by David House 03 Jul '06

03 Jul '06

Hi all. I need a decent regex library and JRegex seems the perfect choice: simple API, yet well-featured, as well as PCRE support. I want to use it on a simple project which involves input files a little larger than typical -- between 100KB and 500KB -- but still small enough so as to not present a problem. However, and I'm fairly sure JRegex is at fault here, my program segfaults on an input of ~230KB. Has anyone used JRegex successfully in this way before? If so, what tactics did you use? Thanks in advance. -- -David House, dmhouse(a)gmail.com

5 6

Input/Output file
by Sara Kenedy 03 Jul '06

03 Jul '06

Hi all, I have a module Test.hs: module Test where import IO readMsgFile = do putStr "Input file: " ifile <- getLine putStr "Output file: " ofile <- getLine s <- readFile ifile writeFile ofile (test s) --test function test :: String -> String test s |s == "true" = "True" |s == "false" = "False" |otherwise = [] Then, I run *Test> readMsgFile Input file: test1.txt Output file: result1.txt The content of test1.txt is string "true". So I expect the result in "result1.txt" is "True" But, open file "result1.txt", it displays empty. I do not know why it does not return correct answer. If you don't mind, please share with me. Thanks a lot. S.

4 3

Re[2]: Debugging with GHCi.
by Bulat Ziganshin 03 Jul '06

03 Jul '06

Hello Pepe, Monday, July 3, 2006, 2:15:59 PM, you wrote: > As you may know, this project was accepted for Google SoC and work started a > few weeks ago. There are now some observable results that can be tested, and > I have set up a wiki page at: > http://haskell.org/haskellwiki/Ghci/Debugger > As we are now approaching the less clear parts of the project your > impressions and ideas (either extense designs or small pointers) are very > welcome. The designs for dynamic breakpoints and call traces are open issues > yet. i'm very pleased with this year's selection of SoC projects and debugging is among the things i'm highly demanded. I personally interested in debugging imperative C-like code, with abilities to set breakpoints, watch (and change?) variable values, see "call stack" (i think that closures stack should serve this need), see local variables at each "call stack" level, evaluate expression and perform IO actions it's especially important for me to be able to use debugger inside some IDE that will allow to set breakpoints and ask variables' values directly inside the source code I'd written. i can recall the following haskell IDEs: HIDE, EclipseFP, Haskell mode for Emacs, Visual Haskell, WinGhci. I think their authors will say better what kind of interface with debugger they prefer imho, just console I/O interface should be enough for most IDEs, as long as debugger's output can be non-ambiguously parsed. Gdb often used to debug C++ programs inside IDEs in this way, i think that cloning it's interface should make it easier to integrate GhcDebugger support into IDEs i also vote for simplest implementation of dynamic breakpoints - the tight integration with IDEs, i.e. _features_, is more important than efficiency (if i correctly understood that this may change only an efficiency) I also will appreciate ability to view call traces - it's the thing that i really need sometimes, especially when debugging unexpected exceptions like "head: empty list". It will be especially great to see all variables' values at each execution moment. I don't have a great ideas about representation of this trace. One obvious idea would be to save some binary file and write special viewer for it, another possible idea is to produce (gigantic) html with trace represented as tree with usual '+' marks to expand each node. ps: you will not get much answers - first, because it's a summer and most Haskellers prefer at this time to set up breakpoints on bodies of their girlfriends ;) second - because this maillist don't have many readers. i suggest you to dup your question into main Haskell list and i cc'ed my answer to haskell-cafe in advance. but nevertheless your project is really important to make Haskell more usable for everyday work and Haskellers returned from their vacations will be glad to see that their debugging tasks becomes much easier. please also write exhaustive documentation at the end of your work so that ghci debugger will not become "black hole" feature -- Best regards, Bulat mailto:Bulat.Ziganshin@gmail.com

1 0