On 13.11.2006 16:54 Valentin Gjorgjioski wrote:

On 13.11.2006 16:48 Pepe Iborra wrote:

...

Hi Valentin

Please, take a look at the Haskell Wiki page for debugging.

http://haskell.org/haskellwiki/Debugging

You will find that thanks to Neil Mitchell there is a Windows version of Hat available. Perhaps you can add your experiences with it if it works for you.

Oh silly me, I was looking that page, but I didn't notice this. Thanks a lot, pepe. I'll try, and I'll post feedback here.

I install it, I tried it, and it doesn't work for me. ./Hat/DML.hs:535:66: Ambiguous type variable `a' in the constraint: `Integral a' arising from use of `*^' at ./Hat/DML.hs:535:66-69 Probable fix: add a type signature that fixes these type variable(s) And this is line 535.... T.uapp2 p125v94v125v148 p125v147v125v147 p (+^) (*^) After that, I have in my code used operator ^, and I deleted that operator. Program finally compiled, but now, hat-observe reports me an error... this program has encountered a problem.... So, I will be really happy with something simple as Hugs.Observe if it works with floats. Some recommendation please? Valentin

On 14/11/2006, at 3:29 AM, Valentin Gjorgjioski wrote:

On 13.11.2006 16:54 Valentin Gjorgjioski wrote:

...

On 13.11.2006 16:48 Pepe Iborra wrote:

...

Hi Valentin

Please, take a look at the Haskell Wiki page for debugging.

http://haskell.org/haskellwiki/Debugging

You will find that thanks to Neil Mitchell there is a Windows version of Hat available. Perhaps you can add your experiences with it if it works for you.

Oh silly me, I was looking that page, but I didn't notice this. Thanks a lot, pepe. I'll try, and I'll post feedback here.

I install it, I tried it, and it doesn't work for me.

./Hat/DML.hs:535:66: Ambiguous type variable `a' in the constraint: `Integral a' arising from use of `*^' at ./Hat/DML.hs:535:66-69 Probable fix: add a type signature that fixes these type variable(s)

And this is line 535.... T.uapp2 p125v94v125v148 p125v147v125v147 p (+^) (*^)

Okay, the type error you are getting is most likely caused by numeric defaulting, or lack of it in Hat. In Haskell, some instances of ambiguous overloading are resolved by the defaulting rules. See: http://haskell.org/onlinereport/decls.html#sect4.3.4 Hat is based on program transformation, and to cut a long story short, the defaulting rules do not work on the transformed program. So you must do the defaulting yourself by adding in explicit type annotations to the offending expression, then transform the program again. In your case it probably amounts to adding something like " :: Integer", or " :: Int" after the expression which uses (^). There is a note about this in the hat faq: http://www.cs.york.ac.uk/fp/hat/faq.html BTW the same problem happens in buddha, which is also based on program transformation. Cheers, Bernie.

On 14.11.2006 23:17 Cale Gibbard wrote:

On 13/11/06, Valentin Gjorgjioski wrote:

...

Following example

import Hugs.Observe

ex8 :: [Float] ex8 = (observe "after reverse" ) reverse [10.0,7.0,3.0,0.0,4.0]

gives me

...

ex8 [4.0,0.0,3.0,7.0,10.0]

...

...

...

...

>>> Observations <<<<<<

after reverse { \ ($-990871 : $-990888 : $-990905 : $-990922 : $-990939 : []) -> $-990939 : $-990922 : $-990905 : $-990888 : $-990871 : [] }

First of all, I don't get this behaviour in Hugs 20050308 on Ubuntu.

Main> ex8 [4.0,0.0,3.0,7.0,10.0]

...

...

...

...

...

>> Observations <<<<<<

after reverse { \ (10.0 : 7.0 : 3.0 : 0.0 : 4.0 : []) -> 4.0 : 0.0 : 3.0 : 7.0 : 10.0 : [] } Just one more thing

If I write ex9 :: [Float] ex9 = (observe "after reverse" ) reverse [10.0,7.0,3.0,0.0,4.0] it doesn't work. If I delete ex9 :: [Float] then it works fine. any suggestions?

On 14/11/06, Cale Gibbard wrote:

On 14/11/06, Valentin Gjorgjioski wrote:

...

Just one more thing

If I write

ex9 :: [Float] ex9 = (observe "after reverse" ) reverse [10.0,7.0,3.0,0.0,4.0]

it doesn't work. If I delete ex9 :: [Float] then it works fine. any suggestions?

This doesn't happen for me. The only thing I can think of trying is to check the type of ex9 by deleting the type signature, loading the file, and typing: :t ex9 on the hugs prompt. If it prints anything other than ex9 :: [Float] then you'll have your answer.

- Cale

Sorry, I just realised what you meant, and I do get the behaviour you initially described without the typesignature there. Actually, without the typesignature, it's inferring the type [Double], due to defaulting. However, adding the typesignature ex9 :: [Double] will again cause it to print the values in the list rather than raw thunks. So perhaps it's because you haven't given a typesignature explicitly, so the list is initially polymorphic, which means that the values in it are actually function calls initially (that is, things like (fromRational 10.0)), and they only end up getting a specialised type later, when hugs finishes compiling the module and applies the monomorphism restriction and defaulting, but hugs isn't going back and optimising the function with the additional knowledge? I don't really know, that's my best guess at it. Maybe someone who knows hugs better would know with more certainty? - Cale

On 15.11.2006 17:38 Ross Paterson wrote:

On Mon, Nov 13, 2006 at 04:32:34PM +0100, Valentin Gjorgjioski wrote:

...

import Hugs.Observe

ex8 :: [Float] ex8 = (observe "after reverse" ) reverse [10.0,7.0,3.0,0.0,4.0]

gives me

...

ex8

[4.0,0.0,3.0,7.0,10.0]

...

...

...

...

>>> Observations <<<<<< >>> after reverse { \ ($-990871 : $-990888 : $-990905 : $-990922 : $-990939 : []) -> $-990939 : $-990922 : $-990905 : $-990888 : $-990871 : [] }

It's a Hugs bug (Float only, Double is fine). Now fixed in CVS.

Oh, thanks a lot. Nice to hear good news.

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

-- Valentin Gjorgjioski Bachelor of Computer Science Department of Knowledge Technologies, Jozef Stefan Institute Jamova 39, SI-1000 Ljubljana, Slovenia Phone: +386 1 477 3343 Fax: +386 1 477 3315 Web: http://kt.ijs.si/ValentinGjorgjioski/ Email: Valentin.Gjorgjioski@ijs.si