[fpc-pascal] Version 1.9.6[??] on the Mac.

[Pedro Lopez-Cabanillas]

On Monday 17 January 2005 08:38, Marco van de Voort wrote:
> smf compiles fine under freeBSD.
Thanks for testing.

> Alsapas has a unit libc dependancy. Most symbols used seem to be simple
> types that are named slightly differently, but the piping functions use
> libc stdio, which is not defined platform independant for FPC. (since FPC's
> stdio doesn't go through libc).
>
> If you don't really use
>
> inputh.inc:function snd_input_stdio_attach(var inputp:Psnd_input_t;
> fp:PIOFILE; _close:longint):longint;cdecl;external asoundlib_name;
> outputh.inc:function snd_output_stdio_attach(var outputp:Psnd_output_t;
> fp:PIOFILE; _close:longint):longint;cdecl;external asoundlib_name;
>
> then it could be made to compile with FPC without using unit libc I thik.

Those functions are seldom used, but are part of the ALSA library API. And the 
main goal for ALSAPAS is to give full support of the ALSA library API. 

Besides inputh.inc and outputh.inc, there are more places where Libc is used.  
See the pollfd references in rawmidih.inc, or timeval/timespec in globalh.inc

But even if you could take away all of them, I don't understand why to do so. 
I mean that ALSA is very Linux specific and it is not possible to use ALSA in 
xBSD, MacOS, Solaris or Win32, so there is not a need to use ALSAPAS outside 
a Linux box. And the ALSA userspace library, libasound.so, depends on Libc 
anyway (and libm, libdl, libpthread...).

> > I've added one more example program, "buildsmf.pas", that creates a MIDI
> > file from scratch, and also fixed several bugs. The ALSA examples still
> > don't run with FPC, because some thread troubles.
>
> buildsmf compiles and generates a 90 byte midifile (under FreeBSD). spysmf
> also compiles. Rest uses ALSA, and doesn't.

OK. Thanks again. I have successfully tested the unit "smf" and the two 
programs buildsmf and spysmf in Win32 today. They should work in any 
environment supported by FPC. The resulting file "test.mid" is 90 bytes 
length, and has a single track with 8 notes, playing a simple C-Major scale. 
The other program should give this output on that file:

$ ./spysmf test.mid
         0      --      SMF Header      Format=0, Tracks=1, Division=120
         0      --      Track           Start
         0       0      Note On         60      120
        60       0      Note Off        60      120
        60       0      Note On         62      120
       120       0      Note Off        62      120
       120       0      Note On         64      120
       180       0      Note Off        64      120
       180       0      Note On         65      120
       240       0      Note Off        65      120
       240       0      Note On         67      120
       300       0      Note Off        67      120
       300       0      Note On         69      120
       360       0      Note Off        69      120
       360       0      Note On         71      120
       420       0      Note Off        71      120
       420       0      Note On         72      120
       480       0      Note Off        72      120
       480      --      Meta Event      End Of Track
       480      --      Track           End

As I have said, the programs involving ALSA are Linux only, but if there is 
some interest on that, I can prepare and distribute a separated package with 
the SMF related files only.

Regards,
Pedro