- Haskell-Cafe - Haskell.org

stateful walk through a tree?
by David Tolpin 19 Feb '07

19 Feb '07

Hi, I am looking for help in design of a stateful tree walker. I have a tree-like data structure (an AST for a small language) for which there is a general stateful walk procedure. That is, each node type is an instance of a walker class, and the walk function is monadic: class Walker a where walk :: a -> State Context b The context is used to store names in the scope, for example. Now, I'd like to use Walker as a general class for implementing several different transformations on the tree (like cross-referencing, code emission, tree visualisation). Those transformations will need expanded state. What is the proper design for that? In Common Lisp, for example, I would Context to contain an opaque slot (visitor), and would assign additional state information to it; a would also define a generic function that would dispatch on the visitor (and probably on walker if I end up having more than one walker). How would I do that in Haskell? I'd like to avoid using mutable variables for now (mostly for didactic puproses). David

2 2

A new code search engine
by Stephane Bortzmeyer 19 Feb '07

19 Feb '07

http://www.krugle.com/ Unlike Google, you can specify Haskell as a language. But the very useful (and, again, unlike Google) parsing service, which allows you to specify that you search "foldl" only in function calls, not function definitions, seems broken for Haskell ("map" is never found in function calls...)

4 3

GHC throws IOError on Win32 when there is no console
by John Ky 19 Feb '07

19 Feb '07

Hi, I noticed on Windows that when I use IO functions that write to stdout when the process is lacking a console, those functions throw an IOError. I'm not sure if this also occurs for stderr because I haven't tried it. Some classes of processes are created without a console because they never interact with the user and include System services. Crashing with IOError in this case is difficult to diagnose because because the only symptom is the process crashes with no visible output. I believe the most sensible behaviour should be for those functions to not throw, but instead do nothing. -John

7 13

questions about core
by Matt Roberts 19 Feb '07

19 Feb '07

Hi all, I am trying to get a deeper understanding of core's role in GHC and the compilation of functional languages in general. So far I have been through - The hackathon videos, - "A transformation-based optimiser for Haskell", - "An External Representation for the GHC Core Language (DRAFT for GHC5.02)", and - "Secrets of the Glasgow Haskell Compiler inliner". and I am still a bit hazy on a few points: - What role do *the semantics* of core play (i.e. how and where are they taken advantage of)? - Exactly what are the operational and denotational semantics of core? - The headline reasons (and any other arguments that emerge) for having core *and* stg as separate definitions. I have an intuition on a few of these points, but I would love something concrete to latch on to. Thanks Matt R

8 12

GHCi and multi-line support ?
by Dunric 19 Feb '07

19 Feb '07

Is it possible to write multi-line definitions in GHC interactive interpreter ? (like in Python interp. ?) When I try to write f.E. if-then-else in more then 1 line I get the following error: <interactive>:1:30: parse error (possibly incorrect indentation) ------------------------------------------ http://icqsms.cz/

3 2

Summarize of Why do I have to specify (Monad m) here again?
by Marc Weber 18 Feb '07

18 Feb '07

Thanks for all the feedback. It did help me a lot. Now I know that if there is something left to discuss the topic should be: Would it make sense to specify partial type declarations ? I don't need an answer right now. Marc

3 2

Re: Re[2]: [Haskell-cafe] Why do I have to specify (Monad m) here again?
by Sebastian Sylvan 18 Feb '07

18 Feb '07

On 2/18/07, Dean Herington <heringtonlacey(a)mindspring.com> wrote: > At 12:42 AM +0400 2/19/07, David Tolpin wrote: > >On Mon, 19 Feb 2007 00:30:47 +0400, Sebastian Sylvan > ><sebastian.sylvan(a)gmail.com> wrote: > >> Well, I guess the H98 report would be a good start. But there are > >> multiple tutorials on type classes that will cover this, most of which > >> are available from haskell.org > > > >Sebastian, > > > >I did read H98 and would like an exact reference. > > See section 4.3.2, the third bullet item in the bulleted list. (Note > that the last sentence of that bullet item says that context > inference--though often possible--is deliberately eschewed.) Yes, but the key point still stands: when the user supplies a type, is should get taken seriously. Just because there may be a way to infer a context which may work in some/most cases, doesn't mean it's a good idea to do so because in all other parts of Haskell the user supplied type is taken seriously -- if you don't include a context, that means there *is* no context. Period. In the original example you really are telling Haskell that you *don't* want any context for "m". So if you want some inference to happen there, you need to find a way to specify that you want "any" contexts rather than no contexts. E.g. instance ( _ m) => GetMV m c where I don't think there's another way to do what you want without seriously messing Haskell up. -- Sebastian Sylvan +46(0)736-818655 UIN: 44640862

1 0

Re: Re[2]: [Haskell-cafe] Why do I have to specify (Monad m) here again?
by Sebastian Sylvan 18 Feb '07

18 Feb '07

On 2/18/07, David Tolpin <dvd(a)davidashen.net> wrote: > >> Why the compiler cannot infer class constraint on m from class > >> definition in instance definition while it can in function type > >> definition? > > > > But it can't! If you give a type to a function, it will assume zero > > class constraints unless you specify them (just like it will when you > > give a type to an instance declaration). If you do something like: > > > > Hi Sebastian, > > it is not the example I brought. In the example I brought I showed how in > function type declaration assertion that an instance of a class is also an > instance of the other class is used. Take a look at my example. According > to what part of the type system logic type inference in instances is not > implemented That's completely different. The class in that case guarantees that the type has an "Eq" class, so it's okay to use the functions in the "Eq" class. You're using the guarantees supplied by the class. When you write instances, it's the other way around, the class has *requirements* that you must fulfill -- and there are multiple ways of doing it (Haskell won't guess, it will obey what you tell it -- if you don't give any class constraints it won't assume that they are there). The point I'm trying to make is that if you say that something is of a given type, Haskell will assume that you know what you're talking about. I.e. if you give a type which has no class constraints, Haskell will assume that you don't want any class constraints -- if this conflicts with something else, you will get an error. I think maybe this is where you're missing the point. *Leaving out* class constraints says something about the type. In other words, if you leave out a class constraint that does *not* mean that you don't care what class constraints the type variable has, it explicitly means that there are *no* class constraints on it (and using it in a context where a class constraint is required, is thus an error). In this case you *require* that any instances of a class is also an instance of the Monad class. But then you try to instantiate a type which is *not* in the monad class in this class. This is a conflict, and you should get an error. If you want to construct a type which is in the Monad class by constraining the type variable "m" to the Monad class, then that's fine, but you have to tell Haskell that this is what you want to do -- if you say that your type has *no* class constraints, it will take your word for it an report the conflict. Another alternative is that you want to produce the type by using a concrete type called, say, "M" which is in the Monad class, then that's also fine. Haskell won't guess what you meant (it can either be a spelling error, "m" for "M", or that you forgot to add "Monad m =>" in the instance declaration, or anynumber of other errors) and change it for you. Whenever you write something out, Haskell will take your word for it, if it fails, it's your fault, and Haskell will tell you what the problem is. -- Sebastian Sylvan +46(0)736-818655 UIN: 44640862

4 6

Very fast loops. Now!
by dons＠cse.unsw.edu.au 18 Feb '07

18 Feb '07

The following C program was described on #haskell #include <stdio.h> int main() { double x = 1.0/3.0; double y = 3.0; int i = 1; for (; i<=1000000000; i++) { x = x*y/3.0; y = x*9.0; } printf("%f\n", x+y); } Which was translated to the following Haskell: {-# OPTIONS -fexcess-precision #-} import Text.Printf main = go (1/3) 3 1 go :: Double -> Double -> Int -> IO () go !x !y !i | i == 1000000000 = printf "%f\n" (x+y) | otherwise = go (x*y/3) (x*9) (i+1) To everyone's surprise, GHC 6.6 beats GCC (3.3.5) here, at least the two test machines: $ ghc -O -fexcess-precision -fbang-patterns -optc-O3 -optc-ffast-math -optc-mfpmath=sse -optc-msse2 A.hs -o a $ time ./a 3.333333 ./a 0.96s user 0.01s system 99% cpu 0.969 total ^^^^^ Versus gcc 3.3.5: $ gcc -O3 -ffast-math -mfpmath=sse -msse2 -std=c99 t.c -o c_loop $ time ./c_loop 3.333333 ./c_loop 1.01s user 0.01s system 97% cpu 1.046 total ^^^^^ Note that newer gcc's will statically compute that loop. Note also that -fexcess-precision must currently be provided as a pragma only. I declare GHC Haskell numerics (with -fexcess-precision) not so shabby! -- Don

11 16

Fwd: Re: Re[2]: [Haskell-cafe] Why do I have to specify (Monad m) here again?
by David Tolpin 18 Feb '07

18 Feb '07

------- Forwarded message ------- From: "David Tolpin" <dvd(a)davidashen.net> To: "Sebastian Sylvan" <sebastian.sylvan(a)gmail.com> Cc: "David House" <dmhouse(a)gmail.com>, haskell-cafe(a)haskell.org Subject: Re: Re[2]: [Haskell-cafe] Why do I have to specify (Monad m) here again? Date: Sun, 18 Feb 2007 23:38:59 +0400 >> Why the compiler cannot infer class constraint on m from class definition in instance definition while it can in function type definition? > > But it can't! If you give a type to a function, it will assume zero > class constraints unless you specify them (just like it will when you > give a type to an instance declaration). If you do something like: > Hi Sebastian, it is not the example I brought. In the example I brought I showed how in function type declaration assertion that an instance of a class is also an instance of the other class is used. Take a look at my example. According to what part of the type system logic type inference in instances is not implemented? David

1 0