[OpenAL] API inconsistency between Linux and Windows

Ed Phillips ed at udel.edu
Wed Nov 2 08:40:18 PST 2005


Hi,

On Tue, 1 Nov 2005, steve wrote:

> Prakash Punnoor wrote:
>> Panne Sven BenQ MD PBM IPM TPS 4 schrieb:
>> 
>>> * Use automake and libtool for the Linux SI and clean up the autoconf 
>>> stuff. This will cause some breakage intially due to the various platforms 
>>> and build variants involved, but these are normal problems during a 
>>> transition phase which can easily and quickly be solved (nightly builds on 
>>> various platforms would help here very much => *hint*). The motivation 
>>> here is that the current build system is not very maintainable and differs 
>>> subtly from what is "usual" in many respects. This should hopefully be 
>>> finished around the end of this month.
>> 
>> I don't know whether this already came up, but what about using cmake
>> (cmake.org) instead of above? In my eyes cmake is much easier to use -
>> and you only have to learn one syntax instead of two (or more).
>> Currently I don't have enough knowledge how to implement every feature
>> done in autoconf/make in OpenAL in cmake, but I guess this will we
>> sorted out.
>
> The problem is that the auto-tools are VERY well known and widely
> adopted in the Linux and BSD communities.  I'd guess that 95 to 98%
> of all projects use them.
>
> Rightly or wrongly, it's a de-facto standard - and I strongly believe
> we should stick with it - although I'm very happy to concede that cmake
> or any of several other tools are easier/better/whatever.

I agree... auto-tools are the way to go.  But the auto-tools can be 
"sucky" if they're just thrown in and allowed to check for "everything 
under the sun" when configure is run.

The OGG and Vorbis libs are particularly painful to build on MinGW/Msys 
(apparently, because of this aspect)... I mean, configure takes 10-15 
minutes to run, but the compile takes under 1-2 minutes.  I don't think 
they need to be checking whether my compiler can cross-compile for PPC 
running in either big-endian or little-endian mode in order to build 
libogg.a on my PC running Windows... ;-P  (That's not much of an 
exaggeration actually...).

Cheers,

 	Ed

Ed Phillips <ed at udel.edu> University of Delaware (302) 831-6082
Systems Programmer III, Network and Systems Services



More information about the Openal mailing list