From rubyforge at lokorin.org  Fri Apr 11 07:02:29 2008
From: rubyforge at lokorin.org (Andreas Launila)
Date: Fri, 11 Apr 2008 13:02:29 +0200
Subject: [gecoder-devel] Simplifying the installation
In-Reply-To: <479881FC.5040207@lokorin.org>
References: <477EB20B.6080502@lokorin.org> <479881FC.5040207@lokorin.org>
Message-ID: <47FF4545.8080609@lokorin.org>

Andreas Launila wrote:
> == Cross-compile Gecode
> 
> This step produces the Gecode DLLs and include files. Those files are
> already included in SVN, so this part should not be needed.
> 
> Download and unpack the Gecode tarball[7] (Gecode/R is currently using
> version 1.3.1). Use the following configuration (possibly with some
> appropriate prefix).
> 
>   ./configure \
>     --with-host-os=Windows \
>     --host=i386-mingw32 \
>     --build=i686-linux \
>     --disable-examples
> 
> Edit Makefile and add "--whole-archive" to DLLFLAGS, then run make.
> 
> A number of DLLs should now have been created, copy them to
> vendor/gecode/win32/lib in the Gecode/R dir. Copy the gecode folder (the
> one containing all the headers) to vendor/gecode/win32/include.
> 

You should also change

  ifeq "no" "yes"
  DLLTARGETS= \

in the Makefile to

  ifeq "yes" "yes"
  DLLTARGETS= \

to build the DLLs. There's probably some easier configuration flag to do
that, but I haven't found it.

-- 
Andreas Launila


From zayenz at gmail.com  Fri Apr 11 07:36:46 2008
From: zayenz at gmail.com (Mikael Zayenz Lagerkvist)
Date: Fri, 11 Apr 2008 13:36:46 +0200
Subject: [gecoder-devel] Simplifying the installation
In-Reply-To: <47FF4545.8080609@lokorin.org>
References: <477EB20B.6080502@lokorin.org> <479881FC.5040207@lokorin.org>
	<47FF4545.8080609@lokorin.org>
Message-ID: <63b5c8b00804110436k53988b8u46332a37e519b39e@mail.gmail.com>

On Fri, Apr 11, 2008 at 1:02 PM, Andreas Launila <rubyforge at lokorin.org> wrote:
> Andreas Launila wrote:

>  You should also change
>
>   ifeq "no" "yes"
>   DLLTARGETS= \
>
>  in the Makefile to
>
>   ifeq "yes" "yes"
>   DLLTARGETS= \
>
>  to build the DLLs. There's probably some easier configuration flag to do
>  that, but I haven't found it.

This is controlled by the --enable-shared switch that configure uses.

Cheers,
Mikael Lagerkvist

-- 
Mikael Zayenz Lagerkvist, http://www.ict.kth.se/~zayenz/

From rubyforge at lokorin.org  Sat Apr 12 08:21:57 2008
From: rubyforge at lokorin.org (Andreas Launila)
Date: Sat, 12 Apr 2008 14:21:57 +0200
Subject: [gecoder-devel] Syntax for extensional constraints
Message-ID: <4800A965.3030305@lokorin.org>

I would like to start adding support for extensional constraints[1]. The
syntax for the tuple variant seems straightforward, but the DFA variant
is not as simple.


== Tuple variant

It seems natural to use a form similar to the domain constraints for
single variables.

  numbers.must_be.in tuple_set

It would also seem natural to let tuple_set be any enumeration of
integers or booleans (depending on whether the left hand side is an
array of integer or boolean variables). The following would constrain
the numbers to be a prime number < 10.

 numbers.must_be.in [2,3,5,7]


== DFA variant

It would probably be easiest to use MiniModel's regular expressions
REG[2] to let the user define regular expressions, which are then
converted to a DFA behind the scenes. I do not think that allowing the
user to specify the DFA directly will add especially much.

A way to make it familiar to Ruby-users would be to have a syntax similar to

  numbers.must.match regexp
  numbers.must =~ regexp

where regexp is a regular expression for integers defined in some way.

The least amount of work would be to more or less expose the REG
operators directly. The following could then be how the regexp "(17 or,
1 followed by 5) any number of times, followed by 4711, repeated 3 to 17
times" would look.

  reg = ((regexp(17) | (regexp(1) + 5))* + 4711)[3,17]

or (using multiple lines)

  tmp = regexp(1) + 5
  reg = (regexp(17) | tmp)*
  reg += 4711
  reg = reg[3,17]

The most natural way would probably be to define it as a normal Ruby
regexp (or just a string), which is then converted back to its source
and reparsed for MiniModel. This will however not work by itself since
the normal Ruby regexp is for characters, while we need a regexp for
integers. I.e. %r{17} is a regexp for character '1' followed by
character '7', while we might have meant the integer 17, or the integer
1 followed by 7. Adding some sort of delimiter to the string
representation could be a possible solution. Using "," as delimiter
would give the following.

  reg = /((17|1,5)*4711){3,17}/

One could also go a bit more verbose but readable route and eschew the
string representation of regexps in favor of something similar to
TextualRegexp[3]. The following could be how the regexp above would look.

  reg = regexp do
    repeat(3..17) do
      at_least_zero do
        any :of do
          integer 17
          group{ integer 1; integer 5 }
        end
      end
      integer 4711
    end
  end

It seems to be a bit too much on the verbose side. TextualRegexp was
however designed for character regexps, so one might be able to slim it
down a bit by using the assumption that only integers are used and that
some functionality isn't included (although one could add sugar for
things such as "maybe").

It would also seem reasonable to have a look at examples where the
constraint is used in practice (such as [4] and [5]) and ensure that
those parts of the examples translate well.


[1] http://www.gecode.org/gecode-doc-latest/group__TaskModelIntExt.html
[2] http://www.gecode.org/gecode-doc-latest/classGecode_1_1REG.html
[3] http://rubyforge.org/projects/texrex/
[4] http://www.gecode.org/gecode-doc-latest/classPentominoes.html
[5] http://www.gecode.org/gecode-doc-latest/classNonogram.html

-- 
Andreas Launila