October 2010 - Haskell-Cafe

ANNOUNCE: Facts
by Alexander Solla 14 Oct '10

14 Oct '10

The Facts hierarchy is meant to contain commonly used, relatively static facts about the "real world". The facts are meant to be encoded using relatively simple Haskell constructs. However, we do make some promises: every data type our modules export will have instances of Data, Eq, Ord, Show, andTypeable. We will use explicit module export lists to control access to internal data structures. As much of the data we are encoding is tabular, we use simple structures like lists and maps to encode the relations. This has two practical ramifications: the textual representation of the data can be very wide, but are also very easy to edit, with "block editing" tools like Vi's visual block mode. The other consequence is that the naive approach to writing queries can be tedious, and the resulting naive queries are slower than they could be. Template Haskell can eliminate much of this drudgery. Felipe Lessa has graciously donated some Template Haskell code which we have adapted. The Facts.* hierarchy currently contains modules with geographical information, such as a data type of countries, cross references to various ISO-3166-1 names for each, a list of states in the United States, and the United States address format. Please see the module hierarchy for more specifics. Patches are welcomed, though prospective contributors are encouraged to encode data structures using lists of pairs to encode bijections, all exposed data types are instances of Data, Eq, Ord, Show, and Typeable, and use explicit exports to only export queries and their atoms. For now, we will add facts to the hierarchy lazily, as our projects need them.

2 2

ANNOUNCE: polyToMonoid
by Kevin Jardine 14 Oct '10

14 Oct '10

Polyvariadic functions that accept an indefinite number of possibly multiple typed arguments are often useful but a bit difficult to declare in Haskell. With some crucial assistance from Oleg Kiselyov, I have created a library that supplies two very general polyvariadic functions that can map their arguments into any monoid you specify. All you need to do to use these is to specify the monoid and some instances of a toMonoid function to map all the types you want to pass as arguments into the monoid. The functions can act as a sort of generalised list, but of course there are also many interesting monoids which are not lists. The library is now up on Hackage here: http://hackage.haskell.org/package/polyToMonoid-0.1 along with detailed documentation and a few examples here: http://hackage.haskell.org/packages/archive/polyToMonoid/0.1/doc/html/Data-… As you can see, the code is short and depends only upon the GHC base library. The original thread discussing the polyToMonoid idea is here: http://groups.google.com/group/haskell-cafe/browse_thread/thread/737c6cf921… Comments and suggestions for improvement are welcome! Kevin

1 0

ANNOUNCE: type-level-natural-number and friends!
by Gregory Crosswhite 14 Oct '10

14 Oct '10

Hey everyone, I am pleased to announce the release of a family of packages for type-level natural numbers. The emphasis on these packages is minimality in order to provide simple core functionality that requires as few extensions as possible beyond Haskell-2010. The (probably foolish) hope is that this will encourage people to use these packages for their simple type-level natural number needs so that the different packages can have interoperable type-level natural numbers. === type-level-natural-numbers v1.1 === This is an update to type-level-natural-numbers, a Haskell-2010 compatible package which provides simple type-level natural numbers. (The only non-Haskell-98 extension that it requires is EmptyDataDecls.) This release adds word aliases for N1..N15 (i.e., One, Two, Three...) as well as a Typeable instance for type-level natural numbers. === type-level-natural-number-operations v1.0 === This package contains the type functions Plus and Minus for performing addition and subtraction on natural numbers; by keeping the functionality minimal I avoided the need to use UndecideableInstances, so the only extension that is required is TypeFamilies. === type-level-natural-number-induction v1.0 === This is the most interesting of the three packages. It provides high-level combinators for expressing inductive operations on data structures tagged with a natural number, which at the moment I call "inductive structures". The idea is that you express your operation in terms of a base case and an inductive case, and then call one of the combinators to do the rest for you. There are combinators for building up an inductive structure from a seed value, folding over/tearing down an inductive structure to get a result, and transforming one inductive structure into another. These combinators are supplied in both monadic and non-monadic versions. The only extension that is required to use this package is Rank2Types. === natural-number v1.0 === This package provides *value*-level natural numbers that are tagged with a type-level natural number corresponding to their value by using GADTs, as well as some simple operations on them. I also provide an instance for the EqT class in type-equality, which means that you can compare two natural numbers with possibly different tags and obtain a proof of equality if and only if they are the same type. The extensions that are required to use this package are Rank2Types and GADTs. (The package itself only uses GADTs, but it pulls in type-level-natural-number-induction which uses Rank2Types.) ======================== Feedback is always welcome! Cheers, Greg

2 4

Installing ghc on NFS ... stripping and upx'ing executables, but what about .a/.so libs
by Ryan Newton 14 Oct '10

14 Oct '10

This gets the ghc executable itself from 42mb down to 5.6mb, which helps a lot on cruddy NFS installations. For example the libHSghc-*, which are pretty huge (49-137mb). Are these libraries that are used only by GHC for compilation? (Actually, I seem to be able to compile simple files without them... so what are they used for?) Any disadvantage to stripping them to save a factor of two? Thanks, -Ryan

1 0

ANNOUNCE: tagged-list v1.0
by Gregory Crosswhite 14 Oct '10

14 Oct '10

Hey everyone, I am pleased to announce the release of tagged-list version 1.0, a package which provides fixed-length lists that are tagged with a phantom type-level natural number corresponding to the length. The advantage of such lists is that you can make static guarantees about them, so that for example the function addLists :: TaggedList n Int -> TaggedList n Int -> TaggedList n Int ... which adds two lists is guaranteed to get two lists of the same length and to produce a list with the same length as the inputs. Some basic operations on these lists have been provided, including instances for Applicable, Traversable, etc. One of the more interesting of these operations is the "join" function", join :: TaggedList m α → TaggedList n α → (TaggedList (Plus m n) α,TaggedList (Plus m n) β → (TaggedList m β,TaggedList n β)) This function takes two lists and appends them, and returns not only the appended list but also a function that takes a list of the same length and splits it into two lists with the same lengths as the inputs; one way of interpreting the second parameter is that it provides a "proof" that a TaggedList of type TaggedList (Plus m n) β can be broken into lists of type TaggedList m β and TaggedList n β. This function is provided because I couldn't figure out how to make a general "split" function pass the type-checker, and fortunately I found that in practice I didn't need a general such function because I was usually taking a list that was the result of some operation applied to two lists I had earlier appended and breaking the result list into sublists the lengths of the original two lists. I also provided functions to convert tagged lists to and from tuples, which can be useful at the times when the type-checker gets confused by your "a :. b :. ... :. E" expression and gives you obscure error messages. Anyway, I hope that the community finds this package useful, and welcome any feedback that you all have to offer. Cheers, Greg

3 4

Books for review in the journal of functional programming
by Simon Thompson 14 Oct '10

14 Oct '10

Just updated: list of books for review in the journal of functional programming, available here: http://www.cs.kent.ac.uk/people/staff/sjt/JFP/available.html Regards Simon T.

1 0

Re: [Haskell-cafe] Client-extensible heterogeneous types (Duck-typed variadic functions?)
by Brandon Moore 14 Oct '10

14 Oct '10

On Oct 13, 2010, at 7:44 PM, Jacek Generowicz <jacek.generowicz(a)cern.ch> wrote: On 2010 Oct 14, at 01:32, Evan Laforge wrote: I think I'm starting too see what my problem is. I think it boils down to hankering for Duck Typing and variadic functions. I fully appreciate that passing functions is a wonderful and powerful technique for catering for variation, but Haskell's type system cramps my style by insisting that I can't put a (Banana -> Cake) in the same container as an (Octopus -> Truffles -> DogsBreakfast). But the thing is, I don't use things like this, even in python. Shame. They're damn useful :-) How are you expecting to call the functions in that container? "for f in c: try: return f(*misc_args) except: pass"? to_do = [(call, (AuntMabel,)), (buy, ([(12*kg, sugar), (6*bushel, wheat)])), (introduce, (Romeo, Juliet))] for do,it in to_do: do(*it) What is the point of doing that? If it's just to defer execution until that loop, you should just rely on lazy evaluation, or [IO ()]. If that's not the only thing you do, then the question is still how you know enough about the structure of values In the list to do anything useful with them. If you just want to be able to inspect them interactively, you should ask about support for inspecting thunks in the ghci debugger. I suppose you haven't heard of parametricity theorems. In Haskell, values have no hair. If you don't know anything about the type of a values you can't inspect it. It's one of the major tools that helps type signatures contribute to the correctness of implementations. In Python, Java, and other similar languages there are lots of things you can do with unknown values - get a string representation, test for equality with another value, get the class it belongs to, etc. So, we won't understand the point of your example without a little more information on what you do to those heterogeneous values, and how the program can tell which things to do wi which item, In Haskell it may be fun to turn on -XGADTs and write data DelayedApp result where Base :: a -> DelayedApp a App :: DelayedApp (a -> b) -> a -> DelayedApp b but it turns out to be isomorphic to data DelayedResult r = DR a Nat - at least until you add some more data to the constructors.

5 8

Re: [Haskell-cafe] Haskellers.com recent changes (and I need some volunteers)
by Michael Snoyman 14 Oct '10

14 Oct '10

Sorry to everyone for not getting back so quickly, I kept getting errors from postfix when I tried sending mail to the cafe. Hopefully this one will go through. As I see it, two open issues are "flagging" and "real Haskellers". Flagging: this was simply a mistake in terminology on my part. I've replaced it with "Report this user." It should be used for either inappropriate content, spam, or someone who's just clearly not part of the community (eg, "Lol, I don't know Haskell, I had Cocoa Puffs for breakfast"). I'm purposely being vague about this; if a user *thinks* there's a problem, it shouldn't take an admin more than a few seconds to investigate it. Now the more important question about real Haskellers: I think I mentioned before implementing the feature that I was a little bit nervous about doing so. The main reason I went ahead and did it anyway was we were getting some... strange gravatars showing up on the homepage. This problem was solved automatically when I added sorting by years of experience (no one can object to Simon PJ and Lennart being on the homepage of course), but the problem with that system is *anyone* can just set their start year to 1990 and get homepage status until an admin blocks him/her. So for the moment, real Haskeller is a minimal whitelisting system, simply intended to prevent people from gaming the system. I've probably chosen bad terminology, and by not explaining this upset a lot of people, my apologies. The point here is not to make the real Haskeller status exclusive, but just to give an extra level of protection. If people really think this is a bad idea, we can take it out. However, keep in mind that the community already seems to favor whitelists (the wiki requires admin intervention for an account, same on HackageDB). This also explains why Lennart and Simon PJ are not at spots 1 and 2: I haven't granted them real Haskeller status. Obviously no offense is intended, I just haven't have a chance yet. As is, I've accepted (virtually) every request for real Haskeller status: the only exception so far has been people who haven't set their full name properly (which once again is my fault for creating a slightly confusing UI). Then the question remains how to we sort people. One idea is randomly; the downside is that it will be difficult to just start browsing the way you can today. We could show newest Haskellers first, which I think definitely has a benefit. I would still recommend we use the real Haskeller status for this, and count people as "new" based on when they receive that status. This will give everyone a chance to make the rounds. And of course, the main feature that's lacking right now is a search/filter for users. I want to wait a few more days to see if we add any more features that should be accounted for in the search. This default sorting issue seems contentious enough that it might be worth doing a poll on it. If you have concrete ideas of how the algorithm should work, please send them over. Make sure to specify whether you think the real Haskeller thing should be kept, and a better name for it if you can think of one. Michael On Mon, Oct 11, 2010 at 3:53 AM, Lennart Augustsson <lennart.augustsson(a)gmail.com> wrote: > The list is clearly not sorted by years of experience. I and Simon PJ > are on the second page, and I doubt many people wrote Haskell programs > before us. > > -- Lennart (iPhone) > > On Oct 10, 2010, at 13:06, Michael Snoyman <michael(a)snoyman.com> wrote: > >> Hey all, >> >> Haskellers became popular a lot faster than I'd anticipated. This has >> prompted me to need to make some changes that I was only planning on >> implementing later on. As usual, all points are up for discussion >> (this is intended to be a community-run site after all). >> >> * Pagination on homepage. We now see only 10 users at a time. >> * The sort order is different. Now, "real haskellers" (we'll discuss >> that in a second) come first, sorted by years of experience with >> Haskell and then alphabetically by full name. >> * We now have site admins. Currently, I'm the only one, but I would >> like to add some more (thus the need for volunteers). Admins currently >> have four rights/duties: >> * Grant/revoke "real haskeller" status. This status currently only >> affects sort order on home page, but might be used for more in the >> future. >> * Grant/revoke admin rights. >> * Block/unblock a user. A blocked user will not show up on the >> homepage, not be counted for public/private/unverified accounts, and >> will have a big nasty "blocked" warning on their user page. >> * Read/close admin messages. These can be created by the following features: >> * Users can request "real haskeller" status in their Edit Profile > Extras tab. >> * Users can request their account to be unblocked. >> * Anyone (logged in or not) can flag a user on their user page. >> >> You'll notice that admins are not able to make any changes to user >> data. Anyone who is interested in taking on this responsibility should >> email me about it. I think five people should easily be able to handle >> this, I'm not anticipating a high load. Then again, I didn't >> anticipate over 200 accounts created in a week either. >> >> Oh, and *please* don't email me personally to request a "real >> Haskeller" status. If you use the button on the site, it makes it much >> easier to add the status. Also, don't take it personally if an admin >> denies your "real haskeller" request; it's currently a very poorly >> defined notion, and I don't even know who the admins are going to be. >> >> Michael >> _______________________________________________ >> Haskell-Cafe mailing list >> Haskell-Cafe(a)haskell.org >> http://www.haskell.org/mailman/listinfo/haskell-cafe >

13 21

Re: [Haskell-cafe] Client-extensible heterogeneous types (Duck-typed variadic functions?)
by Brandon Moore 14 Oct '10

14 Oct '10

If you just want instances of questions you can keep it simple. How about something isomorphic to data Instance = Instance { question : String, answer : String, check : String -> Bool } You could make helper functions similar to your old code, like addition : (Int , Int) -> Instance You might handle problem families by taking a random number generator, or maybe using QuickCheck's Gen monad. easyMultiplication : Gen Instance

2 1

Fuzzy time deltas
by Michael Snoyman 14 Oct '10

14 Oct '10

Hey all, Is there a library that supports fuzzy time deltas? For example, given two UTCTimes (or something like that) it could produce: 43 seconds 13 minutes 17 hours 4 days 8 months I want to use it for the news feature on Haskellers. It's not that hard to write, just wondering if it's already been done. Michael

1 0