April 2004 - Haskell-Cafe

[Haskell-cafe] A useful trick for inclusion and exclusion of commented code
by Sunil KOTHARI 12 Apr '04

12 Apr '04

Here's a little trick that I often use for inclusion and exclusion of a chunk of code. Imagine we have a function foo defined below. The three lines of code can be included or excluded just by a changing a single character --- that's the trick. So, the idea is to comment our code in a slightly different way like this to exclude the code from executing: 1 {- 2 myand :: Bool -> Bool -> Bool 3 myand a b = if a == b then True 4 else False 5 --} Notice how line 5 is commented '--}' instead of '-}'. Now, if we change the first character in line 11 from '{' to '-' the entire chunk of code from line 12-14 no longer remains comment. 11 -- 12 myand :: Bool -> Bool -> Bool 13 myand a b = if a == b then True 14 else False 15 --} This trick can also be applied to C++ comments. Best wishes, Sunil

3 2

Re: [Haskell-cafe] Newbie: Is it possible to catch _|_ ?
by Jon Cast 10 Apr '04

10 Apr '04

Russ Lewis <spamhole-2001-07-16(a)deming-os.org> wrote: > Another newbie question: Is it possible to catch _|_ - that is, to > encounter it, handle it and continue? I'm assuming that you all are > going to say no, since I can't figure out any way to do this and retain > the functional nature of Haskell. This isn't possible in a deterministic language, for several reasons. The IO monad, however, is non-deterministic, and its `catch' function can be used to catch any _|_ value that can be caught, specifically those arising from calls to throw, error, and certain infinite loops. It is non-deterministic, though, so it won't catch all _|_s, nor will it give any guarantee as to which _|_ it will catch (for example, in error 3 + throw (DynException (toDynamic False)) it's indeterminate whether error's return value or throw's return value will ultimately be caught. Jon Cast

5 5

Re: [Haskell-cafe] How to implement `amb' operator?
by Andy Moran 07 Apr '04

07 Apr '04

[ Was sent only to participants, but I meant to send it to the list ] On Wednesday 07 April 2004 06:04 am, Brandon Michael Moore wrote: > Keith is talking about a "comitted choice" style of nondeterminism, where > one of the arguments is picked and the computation continues from there. > > If you want a computation with backtracking, or a list of all possibly > results then you should use the list monad, or another monad that > supports nondetermanism. An important feature of amb is known as "bottom-avoidance", and you can get that with Keith's scheme (although it won't be terribly fair; see mailing list discussions about the fairness or lack thereof of the GHC RTS scheduler), but you can't get it with backtracking (indeed, backtracking is entirely deterministic). Of course, Keith's scheme forces you to work within the IO monad, as well it should. If we allowed amb to have the type amb :: a -> a -> a then all of the transformations carried out within GHC's mid-stage would need to be re-examined for their correctness (in the context of a semantics that understood sharing, as non-determinism makes sharing observable). It's possible that many would still be valid, but it's by no means clear. To my knowledge, no-one has developed a correct semantics for sharing and amb that enables proof of simple transformations like let-floating or lambda-floating or even call-by-need beta reduction, let alone some of the more complex ones like deforestation (which requires a theory that can handle recursion well). Cheers, Andy -- Andy Moran Ph. (503) 626 6616, x113 Galois Connections Inc. Fax. (503) 350 0833 12725 SW Millikan Way, Suite #290 http://www.galois.com Beaverton, OR 97005 moran(a)galois.com

1 0

RE: [Haskell-cafe] How to implement `amb' operator?
by Bayley, Alistair 07 Apr '04

07 Apr '04

Not wanting to appear glib, but... SICP 4.3.3 Implementing the Amb Evaluator http://mitpress.mit.edu/sicp/full-text/book/book-Z-H-28.html#%_sec_4.3.3 Ralf Hinze (2000): Deriving Backtracking Monad Transformers http://www.informatik.uni-bonn.de/~ralf/publications.html#P12 > -----Original Message----- > From: Meihui Fan [mailto:mhfan@ustc.edu] > Sent: 07 April 2004 21:20 > To: haskell-cafe(a)haskell.org > Subject: [Haskell-cafe] How to implement `amb' operator? > > > Hi, > I think it's good idea to compute non-deterministic > problems with the `amb' > operator, just as in LISP/scheme. But how to implement it in haskell? > > noclouds ----------------------------------------- ***************************************************************** Confidentiality Note: The information contained in this message, and any attachments, may contain confidential and/or privileged material. It is intended solely for the person(s) or entity to which it is addressed. Any review, retransmission, dissemination, or taking of any action in reliance upon this information by persons or entities other than the intended recipient(s) is prohibited. If you received this in error, please contact the sender and delete the material from any computer. *****************************************************************

1 0

How to implement `amb' operator?
by Meihui Fan 07 Apr '04

07 Apr '04

Hi, I think it's good idea to compute non-deterministic problems with the `amb' operator, just as in LISP/scheme. But how to implement it in haskell? noclouds

3 2

Newbie: Is it possible to catch _|_ ?
by Russ Lewis 07 Apr '04

07 Apr '04

Another newbie question: Is it possible to catch _|_ - that is, to encounter it, handle it and continue? I'm assuming that you all are going to say no, since I can't figure out any way to do this and retain the functional nature of Haskell.

4 5

Data structure for filesystem representation
by Gerd M 06 Apr '04

06 Apr '04

Hello! For an university project i need to design a data structure (in haskell) that's capable of representing a unix filesystem and therefore fast insertion/deletion. Support for persistent storage is also required. Principally i was thinking to use a Binary Search Tree or an AVL Tree but since this is my first real world application in Haskell I'm not so sure if this is a good (performant) solution. Any advice is welcome! Regards, Gerd. _________________________________________________________________ MSN 8 with e-mail virus protection service: 2 months FREE* http://join.msn.com/?page=features/virus

1 0

RE: [Haskell-cafe] Using -fcontext-stackXX in an OPTIONS pragma
by Simon Peyton-Jones 06 Apr '04

06 Apr '04

Contrary to the manual, -fcontext-stackN is a static flag, so you have to set it on the command line, I'm afraid. I'll fix the manual. Simon | -----Original Message----- | From: haskell-cafe-bounces(a)haskell.org [mailto:haskell-cafe-bounces@haskell.org] On Behalf Of | Anders Höckersten | Sent: 05 April 2004 13:43 | To: haskell-cafe(a)haskell.org | Subject: [Haskell-cafe] Using -fcontext-stackXX in an OPTIONS pragma | | Hi! | I am currently trying to use GHC 6.2 together with the OPTIONS pragma. | My program is generating files and needs to set the context stack | differently for different files. I try to do this using the OPTIONS | pragma, by having this line at the top of my files: | {-# OPTIONS -fglasgow-exts -fcontext-stackXX #-} | (where XX is some generated number) | However, compiling this gives me an error: | Test.hs: unknown flags in {-# OPTIONS #-} pragma: -fcontext-stack60 | | Is it supposed to be possible to set the context stack using the OPTIONS | pragma, and if so, what am I missing? | | /Anders | _______________________________________________ | Haskell-Cafe mailing list | Haskell-Cafe(a)haskell.org | http://www.haskell.org/mailman/listinfo/haskell-cafe

1 0

[Haskell-cafe] Fuse functions given to map and filter without RULES pragma
by Roman Beslik 05 Apr '04

05 Apr '04

I've found a funny way to fuse functions given to map and filter without RULES pragma: toSelect = (,) Just fromSelect = uncurry mapMaybe transformFirst f (x, y) = (f x, y) maybe f = transformFirst (\g x -> g x >>= f) map f = maybe (Just . f) filter f = maybe (\x -> if f x then Just x else Nothing) test :: [Int] test = fromSelect (map (*2) (filter (<10) (map (+3) (toSelect [1,3,4,6,8])))) Everything looks like obvious application of filter and map, but GHC inline (*2), (<10), (+3) (with -O flag) :^). -- Best regards, RB.

1 0

Re: [Haskell-cafe] Context for type parameters of type constructors
by Henning Thielemann 05 Apr '04

05 Apr '04

On 3 Apr 2004, Dylan Thurston wrote: > You might want to add a functional dependency, if you only have one > type of scalars per vertor space: > > > class VectorSpace v a | v -> awhere > > zero :: v > > add :: v -> v -> v > > scale :: a -> v -> v > > But then again, you might not. > > > > instance Num a => VectorSpace a a where > > > zero = 0 > > > add = (+) > > > scale = (*) I see ... functional dependency means "for each 'v' there is only one 'a' that forms a VectorSpace with 'v'". Thus the "VectorSpace a a" instance would exclude any other instance. Since this instance causes 'undecidable instance' and 'overlapping instance' errors in each constellation I tried I finally decided to content myself with special instances instance VectorSpace Double Double instance VectorSpace Float Float These will be the anchors for each type construction recursion needed for more complex VectorSpaces. But e.g. instance VectorSpace (Complex Double) (Complex Double) is still rejected as overlapping with instance (RealFloat v, VectorSpace a v) => VectorSpace a (Complex v) of course ... unfortunately. In addition I separated a class VectorSpacePure from the VectorSpace class that contains the operations that don't need the scalar type 'a', i.e. 'zero' and 'add'. Further I swapped consistently 'a' and 'v' in the class definition and in the quantity data type, i.e. "class VectorSpace a v" and "data Quantity a v". I suppose that it will more oftenly be necessary to say something about the vectors that can built from a certain scalar (i.e. Quantity a) rather than about the scalar types that can be the base of a vector type. > By the way, depending how you resolve the issue above, you might want > instead > > > instance (RealFloat a, VectorSpace b a) => VectorSpace [b] a where > > ... That's really a nice generalization! With the specialised instances for Float and Double as shown above this works. Now I can built VectorSpace over several levels, e.g. a list of complex numbers as vectors with respect to real numbers. But it prohibits functional dependency, right? Functional dependency would probably allow the compiler to detect which scalar type is associated with an vector and thus the compiler wouldn't ask me to tell it explicitly all the time. > Alternatively, if you want to consider varying the scalars, you can > add 'a' as a dummy type variable to 'Quantity': > > > data Quantity v a = Quantity v > > > > instance (Show v, Fractional a, Normed v a) => > > Show (Quantity v a) where > > show (Quantity v) = > > let nv::a = norm v > > in (show (scale (1/nv) v)) ++ "*" ++ > > (show nv) Since I can't use functional dependency I chose this way. Maybe a type class like the VectorSpace finds the way to a revised version of Prelude? Thanks a lot for the help!

2 1