Re: [fluid-dev] [PATCH] Building fluidsynth with MSVC.NET 2003

[Josh Green]

Hello Miguel,

Thank you for the patch contribution.  Unfortunately I have not yet had
time to look at it.  It sounds like it touches on some good things
though.  The extern declaration could be added to the FLUIDSYNTH_API
macro if needed, since that precedes every API function.  If its not
necessary though, I don't see a reason to add it.

The Python binding also sounds really cool.  I'm a big fan of Python and
it would be neat to bundle such a binding with FluidSynth.

I'm working on setting up Trac for the website of FluidSynth.  Hopefully
that will provide a better way for users to contribute.  My goal is to
have that completed by the end of the week.  As usual I've had too
little time to give to FluidSynth, what it really needs and deserves.
Hopefully this will change as the development version of Swami, which is
based on FluidSynth, nears a releasable stage.

Cheers!

Josh

On Sat, 2007-04-07 at 16:43 +0200, Miguel Lobo wrote:
> Hi list,
> 
> Short story: please find attached a patch against current CVS that
> fixes various problems with the Win32 port, including:
> * Execution bugs in the standalone executable and static library.
> * A number of compiler errors and warnings with MSVC.NET.
> 
> Long story: a while ago I wrote to this list about my intention to use
> fluidsynth in a Linux/Windows pseudo-soundtracker I'm writing in
> Python.  The GUI is still in a very early state, but I thought I'd
> start writing the fluidsynth Python bindings so I can find any
> fundamental problems with my approach before it's too late. 
> 
> Well, the first snag I've hit so far is compiling fluidsynth using the
> free MSVC.NET 2003 compiler.  I'm using this compiler because it's the
> same one the Python project uses for its Windows release and that
> makes things easier for me. 
> 
> Some of the problems I've fixed or worked around are:
> * autotools doesn't seem to support MSVC.  I solved this by using
> QMake (I have not included the QMake project file in my patch, but I
> can provide it if someone is interested). 
> * Some files included config.h without checking for HAVE_CONFIG_H;
> this does not work if you are not using the autotools.
> * Part of the initialization necessary for the DirectSound driver had
> been moved to DllMain, so it was only executed in the DLL.  I've moved
> this back to the DirectSound driver. 
> * Command line argument processing was messed up if you don't have
> getopt; I hope my simple fix does not break anything.
> 
> I've also renamed fluid_dll.c to fluid_win32.c (as the code it
> contains is needed in Win32 even if you are not compiling a DLL), and
> created a fluid_win32.h header. 
> 
> Finally, the biggest part of the patch actually consists of adding
> "extern" in front of all the function prototypes in the headers.  I
> did this because I mistakenly thought it was the cause of some
> compiler and linker warnings I was seeing.  Hopefully keeping those
> "extern" is a good idea anyway. 
> 
> With this patch, the standalone executable and DLL build with MSVC.NET
> without any warnings, and I've also checked fluidsynth is able to play
> MIDI files. 
> 
> Perhaps Josh or somebody can check my patch does not break the Linux
> build and then check it in the repository. 
> 
> Thanks,
> Miguel
> 
> _______________________________________________
> fluid-dev mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/fluid-dev