- Libraries - Haskell.org

Proposal: Add --with-libedit flag to the readline package
by Judah Jacobson 14 Jan '08

14 Jan '08

It would be useful for the readline package to support compiling against the libedit library (which provides a subset of the readline APIs): - libedit is available by default on OS X. - since libedit is BSD-licensed, there are no problems statically linking ghc with it. (This may be useful on Windows.) I propose adding a --with-libedit flag to the readline autoconf script. Without that flag, the package will behave exactly as before, refusing to link against libedit. With that flag, the following behavior occurs: - GNUreadline.framework (OS X - only) is ignored, if present - We try to link with -lreadline, and don't fail if readline is actually libedit. - If it is libedit, we #ifdef out all of the functions not supported by libedit. (these are generally low-level APIs not needed by most applications, including ghci.) Otherwise, if we're linking against GNU readline, we support all the available APIs. Suggested deadline: Jan. 23, 2008. A patch is attached to the trac ticket: http://hackage.haskell.org/trac/ghc/ticket/2028 Thoughts? -Judah

5 11

Copyright for readline/editline packages
by Judah Jacobson 12 Jan '08

12 Jan '08

Hi all, As mentioned before, I'm putting together an editline package which will provide a subset of the APIs from the readline package, but be BSD-compatible since it links against libedit. The licensing for the readline package itself is a little strange -- it's licensed under the GPL (because it links with libreadline), but there's no copyright holder information; and Readline.hsc from that package lists "Copyright: (c) unknown". I'd like to put the Haskell editline package under BSD3, but most of the code will be copied verbatim from the readline package. How should this be attributed/copyrighted? Thanks, -Judah

2 1

PROPOSAL: Some more 'Applicative' combinators
by Iavor Diatchki 11 Jan '08

11 Jan '08

Hello, I propose that we add the following combinators to the 'Control.Applicative' module: skipMany :: (Alternative f) => f a -> f () skipMany p = skipSome p <|> pure () skipSome :: (Alternative f) => f a -> f () skipSome p = p *> skipMany p endBy :: (Alternative f) => f a -> f b -> f [a] endBy p s = many (p <* s) endBy1 :: (Alternative f) => f a -> f b -> f [a] endBy1 p s = some (p <* s) sepBy :: (Alternative f) => f a -> f v -> f [a] sepBy p s = sepBy1 p s <|> pure [] sepBy1 :: (Alternative f) => f a -> f v -> f [a] sepBy1 p s = (:) <$> p <*> many (s *> p) Any objections? Deadline for discussion is 2 weeks from now, which would be the 20th of Jan. -Iavor

7 10

Proposal: new version of getModificationTime
by Felix Martini 11 Jan '08

11 Jan '08

Hello all, Function getModificationTime in System.Directory returns ClockTime. ClockTime is part of the old-time package. The (new) time package has UTCTime, so i propose a new version of getModificationTime that returns UTCTime instead. I am not sure what a good name for the new version would be, perhaps getModificationUTCTime? Regards, Felix

2 3

Announcement: Typeful [x]html combinators -- pre-release 0
by Jon Fairbairn 11 Jan '08

11 Jan '08

... or, "whoops! I've writtten another html combinator library." History: I was surprised to find that all the Haskell html generating stuff I've tried allowed one to construct invalid HTML. I thought to myself that this might make an undergraduate project, but then I thought "maybe it's too hard? I'd better write some of it to see how hard it is." After a while I decided that it was probably a bit too big for an undergraduate project, but by that time I'd written too much of it to throw away, and just had to keep going so here it is. I'm announcing it here because I hope the audience is fairly small but discerning, and I'm announcing it at all because I've run out of steam for the moment. This is the first code above the size of a nonce-programme that I've written since I got ill years ago, so don't go too hard on me! I still am ill, so don't expect swift responses either. I'm mainly interested in opinions on the questions below, and in test cases that fail validation. You can get it with darcs get --partial http://homepage.ntlworld.com/jon.fairbairn/Typeful/Text/HTMLs and the documentation (such as it is) is at <http://homepage.ntlworld.com/jon.fairbairn/Typeful/Text/HTMLs/Documentation…> Here's the announcement (which might better be called the release note, except that I'm not sure this counts as a relase). ------ Typeful HTMLs library Library of types and combinators to produce valid xhtml 1.0 (and eventually html 4.01) -- if you use this library (and if it's not faulty), you won't need to run the validator over the output of your code. Types prevent invalid nesting of elements (including the prohibitions of appendix B of the xhtml standard) and ensure an attribute is only passed to elements that has it. Most attributes are given types that reflect their content rather than just being strings (although some are just wrappers around strings for the moment). Includes full set of xhtml character entity names Includes type with names for all IANA-registered character sets. Provides the option of generating output in a choice of character sets (currently with the wide choice of utf_8, us_ascii, latin1 or iso_8859_5 (cyrillic)). Provides the option of generating xhtml without an xml version declaration (to avoid putting IE into quirks mode). It's generated by TemplateHaskell, but otherwise Haskell98. Questions: * I want to add a function to produce HTML 4.01. Unfortunately there are trivial differences (body takes a nonempty list in HTML but an ordinary list in xhtml, and similar). One approach (and the one I'd choose if left to my own devices) would be to restrict the document tree to the common subset, so the types would be constrained by whichever language was the more restrictive in any given case. An alternative would be to transform the xhtml tree by adding empty div elements. Which do people prefer? * Having a monad inside the result types of the *_allowed_in classes complicates the types, complicates programmatic generation of elements and makes some error messages much worse. It wouldn't be necessary if we simply wrote Haskell lists instead of using the +++ operator, ie instead of body << (h1 << ... +++ p << ... +++ p << ...) we had to write body << [h1 << ..., p << ..., p << ...] Admittedly, it is rather more awkward for nonempty lists: ul << (li <<... +++ li<< ... +++ li<<...) would have to be written ul << (li<<... +:[li<<...,li<<...]). Would anyone really be bothered by that? (Either that, or can someone come up with a way of rearranging the types so that +++ works without the monad in the result types of elements?) * What should I do about REQUIRED attributes? Some alternatives are listed in the TODO file. * The preference carrying monad used for Render is backwards; I tried making it go forwards, but the Laziness test programme took about 1.7 times as long to run. I'd like to put some state in there, which can't be done backwards. Can I do this somehow without the performance hit? * This pre release has a very restrictive license (barely falling short of "You may not use this!") My preference is, I think, to release it under LGPL. Is that going to cause people undue grief? -- Jón Fairbairn Jon.Fairbairn(a)cl.cam.ac.uk

5 14

Add 'subsequences' and 'permutations' to Data.List (ticket #1990)
by Twan van Laarhoven 10 Jan '08

10 Jan '08

Hello all, As Thomas Hartman noted in a recent cafe thread [1], haskell 1.3 included the functions 'subsequences' and 'permutations'. I think these functions are quite useful, and I don't know why they were ever removed. This is a proposal to add these two functions to Data.List. The implementation is taken directly from the Haskell 1.3 report [2]. Trac ticket: #1990 Discussion period ends: January 7th (since there is a holiday comming up) Twan [1] http://article.gmane.org/gmane.comp.lang.haskell.cafe/33535 [2] http://haskell.cs.yale.edu/haskell-report/List.html

8 21

Tests in Cabal
by Adrian Hey 10 Jan '08

10 Jan '08

Hello, I've been playing about with the option to add tests to packages and have modified the Setup.hs as follows.. import Distribution.Simple import MyPackage main :: IO () main = defaultMainWithHooks (simpleUserHooks {runTests = myTests}) myTests :: Args -> Bool -> PackageDescription -> LocalBuildInfo -> IO () myTests _ _ _ _ = <whatever> Which I think is the right thing to do if I understand the docs. The first problem I come up against is that in order to do this practically all the modules from the package need compiling by runghc (even during the configure stage presumably). As I guess most people will be using runHaskell/runghc to do this it seems like this will all take quite a bit of time. In my case the process fails completely because the package library files need CPPing and is seems runghc doesn't do this (by default at least). Is this really the way tests are supposed to be included? It doesn't seem workable to me. I wonder if it would be better to just build the lib first (without tests) and then separately build one or more test executables? What does everyone else do about this? Thanks -- Adrian Hey

3 3

typo in Data.Binary documentation
by Chad Scherrer 09 Jan '08

09 Jan '08

The latest Data.Binary documentation says instance of Binary should have get . put == id But this doesn't typecheck! I'm guessing it's probably supposed to say decode . encode == id -Chad

4 3

[Fwd: Re: Maps]
by Christian Maeder 09 Jan '08

09 Jan '08

-------- Original Message -------- Subject: Re: Maps Date: Tue, 08 Jan 2008 05:35:07 -0800 From: Mike Hamburg <mike(a)mikehamburg.com> To: Christian Maeder <Christian.Maeder(a)dfki.de> CC: paul(a)cogito.org.uk References: <47832E64.3090008(a)dfki.de> So here's an approximation of my problem. I have a map of objects, possibly aggregated from different sources. They have (in a case of medium complexity) type (t,u) where the t represents the source and u represents the position within that source. They're derived from something like a Map t (Map u v), but it is convenient for other reasons to represent this as an actual Map (t,u) v. (It's possible that I'll end up hacking around this with some type class or another, but let's ignore that possibility for now.) I need to efficiently represent ranges of objects to be viewed or deleted from this map, in such a way that the map library can extract or delete them quickly. But I might have, say, only a t value, and no values of u available. On Tue, 2008-01-08 at 09:03 +0100, Christian Maeder wrote: > I'm not quite sure what you are asking for here. Why does the first > argument take both a key and a value, and what is the Ordering saying? > >From your description I was expecting something more like: > > interval :: k -> k -> Map k v -> Map k v > > Are you saying that if the Ordering is "EQ" then the key is within the > range? Yes. The "ordering" produced is a sort of disjunctive ordering: it's either LT all, or GT all, or EQ at least one element in the interval. I suppose my version should really have type (k -> Ordering) -> Map k v -> Map k v because the values shouldn't come into play unless they're monotonically dependent on the keys. > Take a look at my Ranged Sets library (on Hackage) for a complete > treatment of the concept of ranges within an ordered type. I can > imagine something like > > rangedSetMap :: RSet k -> Map k v -> Map k v Something like an RSet, with the following operations: union, singleton, Range -> RangeSet, ... upcast :: RangeSet t -> RangeSet (t,u) -- and similar for other product types intersection, difference :: Map k v -> RangeSet k -> Map k v > where the resulting Map only contains the keys that are in the RSet. At > present you would have to do that by iterating through the Map, but I > can imagine a more efficient method. Alternatively the Boundary type > defined in that library might be used: it splits an ordered type into > values above and below a boundary. (Two Boundary values make a Range, > and an ordered list of non-overlapping Ranges is an RSet). Yes. But I think what you need to efficiently implement this is some sort of implicit cut, of the form (k -> Something) -> Map k v -> (Map k v, Map k v). > Paul. Also, I'm sort of curious: in C/C++/Java/etc, you can produce a Dense type which has: instance (Eq, Ord, Bounded) Dense between :: Dense -> Dense -> Dense -- if a < b then a < between a b < b This is sort of a lie: between is not referentially transparent, in that two calls to (between a b) may produce results that aren't ==. Still, the efficiency is impressive: ==, >, <, etc are O(1) 'between' is O(log n) where n Dense's are currently live. You can actually make it O(1), but the constant is big, so people don't. total space usage is O(n) integers of 2(log n) bits each Is there any way to even approach this in Haskell? I mean, you can use [Int], but that's not nearly so fast... Mike

3 2

ANN: Build fixed for regex-base,posix,compat,pcre
by ChrisK 08 Jan '08

08 Jan '08

ANNOUNCEMENT: Build fixed for regex-base, regex-posix, regex-compat, regex-pcre The changes are mainly to the Cabal build files to support ghc-6.8 and ghc-6.6 simultaneously. They definitely work with cabal version 1.2.3.0 (required for regex-pcre). The regex-base, regex-posix, and regex-compat packages also seem to work with cabal-1.2.2.0 (but not regex-pcre). The darcs repositories under http://darcs.haskell.org/packages/regex-unstable/ hold the new code. I have uploaded new versions of the most popular regex-* packages to hackage: Mon Jan 7 13:18:00 PST 2008 ChrisKuklewicz regex-pcre-0.94.1 Mon Jan 7 13:05:57 PST 2008 ChrisKuklewicz regex-posix-0.93.1 Mon Jan 7 13:04:49 PST 2008 ChrisKuklewicz regex-base-0.93.1 Mon Jan 7 06:50:54 PST 2008 ChrisKuklewicz regex-compat-0.91 I would like to thank everyone who sent me updated, and particularly Olivier Boudry who sent patches. These build fixes have only been tested with ghc-6.6.1 on Mac OS X 10.5.1 (powerpc) and with ghc-6.8.2 on Linux (x86). Happy New Year, Chris Kuklewicz

1 0