Support for detection of SWIG and MONO during configuration

[Daniel Morissette]

I have created bug 1720 for the generic SWIG detection and as a tracking 
bug for language-specific bugs:
http://mapserver.gis.umn.edu/bugs/show_bug.cgi?id=1720

To keep things in order please do not include language-specific details 
in this bug, instead create new bugs for each MapScript language that we 
want to detect, assigned to the right person for each language and mark 
them as blocker for this bug.

Daniel


Daniel Morissette wrote:
> I agree with Howard that the best place to start would be to document 
> the requirements for each MapScript flavour in separate bugs, then once 
> we have sorted out the details I could try to help with the configure.in 
> updates.
> 
> A few questions to address:
> 
> 1- How to detect SWIG, is it enough to look for the swig executable and 
> test its version? Also which version is required as a minimum?
> 
> 2- Do we want the main Makefile automatically build all the enabled SWIG 
> bindings the same way that PHP MapScript is currently built 
> automatically by the main Makefile when it is enabled? I would think the 
> answer is yes, but we need to agree on that.
> 
> 3- Then for each supported language, we need:
>     - Makefile template, including targets for default build and for 
> install (optionally a 'make test' too)
>     - List of dependencies to detect (e.g. /usr/bin/mcs for MONO?)
>     - Required version(s) of dependencies
> 
> Daniel
> 
> 
> Sean Gillies wrote:
> 
>> At the very least, this could banish forever the mapscriptvars file  
>> hack.
>>
>> Sean
>>
>> On Mar 21, 2006, at 7:58 AM, Howard Butler wrote:
>>
>>> I also think it would be nice to detect SWIG and the SWIG MapScript
>>> variants within configure for the next MapServer release.  We already
>>> allow the specification of PHP in configure.  Some complications
>>> might be that certain MapScripts may have specific minimum SWIG
>>> requirements, and some MapScripts are not built with a normal
>>> GNUMakefile (Python uses distutils, for example).  Doing this might
>>> also make packager's lives easier.
>>>
>>> A bug with the specifics for the detection and building of each
>>> MapScript would be a good place to start.  I can help with the Python
>>> side.
>>>
>>> Howard
>>>
>>> At 3:46 AM -0600 3/21/06, Tamas Szekeres wrote:
>>>
>>>> Daniel,
>>>>
>>>> According to the IRC conversation I've posted this message to  
>>>> review it more
>>>> publicly.
>>>>
>>>> The UNIX build would be more uniform if we could make support for  
>>>> detecting
>>>> SWIG and MONO at the ./configure phase.
>>>> I would suppose to have optional features should be given as a  
>>>> parameter like
>>>>
>>>> ./configure --with-swig --with-csharp
>>>>
>>>> by allowing to add the path to the swig and the csharp compiler  
>>>> executables.
>>>> I think for swig /usr/local/bin/swig and /usr/bin/swig
>>>> for mono
>>>> /usr/local/bin/mcs and /usr/bin/mcs should be searched for by  deafault
>>>>
>>>> The other languages may have other requirements.
>>>>
>>>> Any other suggestions would be appreciated.
>>>>
>>>> Thanks,
>>>>
>>>> Tamas
>>
>>
>>
> 
> 


-- 
Daniel Morissette
http://www.mapgears.com/