From metawilm at gmail.com  Sat Mar 15 17:09:05 2008
From: metawilm at gmail.com (Willem Broekema)
Date: Sat, 15 Mar 2008 18:09:05 +0100
Subject: [gecol-devel] Gecol 2.0 fixes for Allegro
Message-ID: <f6bc9b490803151009y24216febv5b4849e223126661@mail.gmail.com>

Hello,

Being interested in contraint programming, I had a look at Gecode
using Gecol. Having Lisp as interface language is great!
Here are a few patches to make the code in the "gecol20" branch
compile, and the examples work, in Allegro 8.1.

 - "allegro" fixes warnings for top-level export calls, and a symbol
case issue in Allegro's Modern mode
 - "allegro_macosx" fixes the gcc options, to look for both :darwin
and :macosx in *features* for dynamiclib. (Allegro only includes
:macosx)
 - "gecol_space_calls" fixes compile errors in the ffi calls (taking
Gecode release-2.1.1 as target), though not sure this is the best way
to solve it (where should the matching delete's be?)

This is the first time I'm sending Darcs patches, so let me know if
there's anything not quite right.

Cheers,
- Willem
-------------- next part --------------
A non-text attachment was scrubbed...
Name: allegro_macosx.patch
Type: text/x-patch
Size: 715 bytes
Desc: not available
URL: <https://mailman.common-lisp.net/pipermail/gecol-devel/attachments/20080315/c54b8c22/attachment.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: gecol_space_calls.patch
Type: text/x-patch
Size: 1120 bytes
Desc: not available
URL: <https://mailman.common-lisp.net/pipermail/gecol-devel/attachments/20080315/c54b8c22/attachment-0001.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: allegro.patch
Type: text/x-patch
Size: 949 bytes
Desc: not available
URL: <https://mailman.common-lisp.net/pipermail/gecol-devel/attachments/20080315/c54b8c22/attachment-0002.bin>

From kilian.sprotte at googlemail.com  Wed Mar 19 09:38:07 2008
From: kilian.sprotte at googlemail.com (Kilian Sprotte)
Date: Wed, 19 Mar 2008 10:38:07 +0100
Subject: [gecol-devel] Gecol 2.0 fixes for Allegro
In-Reply-To: <f6bc9b490803151009y24216febv5b4849e223126661@mail.gmail.com>
References: <f6bc9b490803151009y24216febv5b4849e223126661@mail.gmail.com>
Message-ID: <1d26dc7e0803190238h65d3e43m49fd3847108774fc@mail.gmail.com>

Hi Willem,

thank you very much for helping with Allegro and the newest GECODE version.
All the patches arrived fine!
I have now pushed them to gecol20 - but committed a small change
afterwards...

You made this change:

hunk ./gecol.asd 64
-  (let ((loader (intern "LOAD-FOREIGN-LIBRARY" :cffi)))
+  (let ((loader (intern '#:load-foreign-library :cffi)))
}

For instance SBCL would complain if the first arg of intern is not a string
(and according to the spec it should not be anything else). As I dont have
Allegro, what is the exact error message / problem here? Maybe we could use
find-symbol with a better chance for all lisps? - For the moment, I have
changed it back to a string, sorry.

Then there were these questions:
hunk ./glue.cpp 564
+  // WB: delete opt?
hunk ./glue.cpp 1450
-  return set->variable();
+  // WB: was: return set->variable();
+  return set->var();
}

Right now, I am a bit out of business with gecol, as I am currently not
using it and working on other things. Maybe Mauricio can help with this?

Thanks again for your help,
best,
  Kilian

On Sat, Mar 15, 2008 at 6:09 PM, Willem Broekema <metawilm at gmail.com> wrote:

> Hello,
>
> Being interested in contraint programming, I had a look at Gecode
> using Gecol. Having Lisp as interface language is great!
> Here are a few patches to make the code in the "gecol20" branch
> compile, and the examples work, in Allegro 8.1.
>
>  - "allegro" fixes warnings for top-level export calls, and a symbol
> case issue in Allegro's Modern mode
>  - "allegro_macosx" fixes the gcc options, to look for both :darwin
> and :macosx in *features* for dynamiclib. (Allegro only includes
> :macosx)
>  - "gecol_space_calls" fixes compile errors in the ffi calls (taking
> Gecode release-2.1.1 as target), though not sure this is the best way
> to solve it (where should the matching delete's be?)
>
> This is the first time I'm sending Darcs patches, so let me know if
> there's anything not quite right.
>
> Cheers,
> - Willem
>
> _______________________________________________
> gecol-devel mailing list
> gecol-devel at common-lisp.net
> http://common-lisp.net/cgi-bin/mailman/listinfo/gecol-devel
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.common-lisp.net/pipermail/gecol-devel/attachments/20080319/924a5d7a/attachment.html>

From mauriciotorob at gmail.com  Wed Mar 19 11:54:10 2008
From: mauriciotorob at gmail.com (Mauricio Toro)
Date: Wed, 19 Mar 2008 12:54:10 +0100
Subject: [gecol-devel] Welcome Willem
Message-ID: <a6b4b44a0803190454o4414b6b2k9ab20613b8ce95ca@mail.gmail.com>

Hello,

I am taking care of Gecol these days.
Since Kilian is working on something else.

I guess the changes on the lisp files are fine.
I am going to check the ones in the .cpp
with more detail and I will tell you later.

By the way, is great that you want to help.
Right now, I have 2 things in mind:

1) Give support to the RegExp class and the DFA class
from Gecode in Gecol. Probably making functions
for all methods, because using Lisp classes can be messy.

2) Give more support to sets, I made some examples
but I think it is required to make something like
what it was done with with-var-args for intVarArrays
in the setVarArrays.

Please contact me if you want to colaborate with some
of that. You can also contact the Gecode team if you have
specific questions about Gecode, they are helping
with us.

Mauricio

-- 
Mauricio Toro Bermudez
Estudiante de Ingenier?a de Sistemas
Pontificia Universidad Javeriana, Colombia

Stagiare ? l'Ircam
1, place Igor-Stravinsky 75004 Paris,
France de 2008 ? 2009


From metawilm at gmail.com  Sat Mar 22 20:55:22 2008
From: metawilm at gmail.com (Willem Broekema)
Date: Sat, 22 Mar 2008 21:55:22 +0100
Subject: [gecol-devel] Gecol 2.0 fixes for Allegro
In-Reply-To: <1d26dc7e0803190238h65d3e43m49fd3847108774fc@mail.gmail.com>
References: <f6bc9b490803151009y24216febv5b4849e223126661@mail.gmail.com>
	<1d26dc7e0803190238h65d3e43m49fd3847108774fc@mail.gmail.com>
Message-ID: <f6bc9b490803221355x52830c21oc0a6203388188136@mail.gmail.com>

On Wed, Mar 19, 2008 at 10:38 AM, Kilian Sprotte
<kilian.sprotte at googlemail.com> wrote:
> thank you very much for helping with Allegro and the newest GECODE version.
> All the patches arrived fine!
> I have now pushed them to gecol20 - but committed a small change
> afterwards...
>
> You made this change:
>
> hunk ./gecol.asd 64
> -  (let ((loader (intern "LOAD-FOREIGN-LIBRARY" :cffi)))
> +  (let ((loader (intern '#:load-foreign-library :cffi)))
> }
>
> For instance SBCL would complain if the first arg of intern is not a string
> (and according to the spec it should not be anything else). As I dont have
> Allegro, what is the exact error message / problem here?

The problem is that in a case-sensitive Lisp (Allegro's "Modern"
mode), the symbol is exported by cffi as lowercase. (intern
"LOAD-FOREIGN-LIBRARY" :cffi) creates a second, different symbol. To
solve this, the argument to intern (or find-symbol) should be given in
lowercase.

Allegro happens to also accept symbols as argument to intern, so I was
not notified of the mistake (it is controlled by the switch
excl:*intern-allows-symbol*, by default true).

Now I'd like to propose (intern (string '#:load-foreign-library)
:cffi), to fix the case issue.

Cheers,
- Willem


From kilian.sprotte at googlemail.com  Mon Mar 24 19:04:27 2008
From: kilian.sprotte at googlemail.com (Kilian Sprotte)
Date: Mon, 24 Mar 2008 20:04:27 +0100
Subject: [gecol-devel] Gecol 2.0 fixes for Allegro
In-Reply-To: <f6bc9b490803221355x52830c21oc0a6203388188136@mail.gmail.com>
References: <f6bc9b490803151009y24216febv5b4849e223126661@mail.gmail.com>
	<1d26dc7e0803190238h65d3e43m49fd3847108774fc@mail.gmail.com>
	<f6bc9b490803221355x52830c21oc0a6203388188136@mail.gmail.com>
Message-ID: <1d26dc7e0803241204p24bc1354y3e726607d71b8312@mail.gmail.com>

Hi Willem,

>  Now I'd like to propose (intern (string '#:load-foreign-library)
>  :cffi), to fix the case issue.

ok, now I understand, why you are insisting to read a symbol.
It's pushed now.

Thanks!
  Kilian


From mauriciotorob at gmail.com  Tue Mar 25 17:08:13 2008
From: mauriciotorob at gmail.com (Mauricio Toro)
Date: Tue, 25 Mar 2008 18:08:13 +0100
Subject: [gecol-devel] About Willem's changes
Message-ID: <a6b4b44a0803251008t2e3f9041x9506add005ec540d@mail.gmail.com>

Hello Willem,

I could not compile the gecol package
with the changes you  have made.
I am getting those gcc errors:
- Options is not in the Search namespace.
- var is not a member of VarSet

Any clues why?

-- 
Mauricio Toro Bermudez
Estudiante de Ingenier?a de Sistemas
Pontificia Universidad Javeriana, Colombia

Stagiare ? l'Ircam
1, place Igor-Stravinsky 75004 Paris,
France de 2008 ? 2009


From metawilm at gmail.com  Tue Mar 25 17:50:28 2008
From: metawilm at gmail.com (Willem Broekema)
Date: Tue, 25 Mar 2008 18:50:28 +0100
Subject: [gecol-devel] About Willem's changes
In-Reply-To: <a6b4b44a0803251008t2e3f9041x9506add005ec540d@mail.gmail.com>
References: <a6b4b44a0803251008t2e3f9041x9506add005ec540d@mail.gmail.com>
Message-ID: <f6bc9b490803251050s5a33df8cs29a343486cad41a@mail.gmail.com>

Mauricio,

On Tue, Mar 25, 2008 at 6:08 PM, Mauricio Toro <mauriciotorob at gmail.com> wrote:
>  I could not compile the gecol package
>  with the changes you  have made.
>  I am getting those gcc errors:
>  - Options is not in the Search namespace.
>  - var is not a member of VarSet
>
>  Any clues why?

Could the difference be the Gecode version with which you are interfacing?
My changes made things work with release-2.1.1 (as I indicated in my message).

- Willem