Default audio output method should be taken from real world, not from outer space :)

[Bohdan R . Rau]

Hi

Simple but annoying problem - default audio output method.

In current code we have very interesting fragment (configuration.c):

#ifdef __SUNPRO_C
/* Added by Willie Walker - default to OSS for Solaris */
    GlobalFDSet.audio_output_method = g_strdup("oss");
#else
    GlobalFDSet.audio_output_method = g_strdup("pulse");
#endif /* __SUNPRO_C */

Of course it works if we compile speech-dispatcher for Solaris or some
Linux distribution (like Ubuntu) which has pulse compiled.

On my system (PLD Linux) I have no pulse (and I don't need pulse, and even
I don't want to hear anything about pulse on my box). I configure
speech-dispatcher with only alsa as audio subsystem. But still default
audio subsystem is pulse, not alsa, and I must manually configure alsa in
speechd.conf after each compilation...

For me it's big, old and ugly bug.

How it should work?

Let's assume we have some AC_DEFINE macros in configure.in - for each
really compiled audio subsystem there should be corresponding definition in
config.h file. In this case we could define default output method as some
really compiled in - for example, if we don't compile pulse, the "pulse"
should not be default audio! It should be declared as macro depending on
some HAVE_AUDIO_something definitions in config.h.

Something like:

#ifdef HAVE_AUDIO_PULSE
#define DEFAULT_AUDIO_METHOD "pulse"
#else
... define other method here
#endif
GlobalFDSet.audio_output_method = g_strdup(DEFAULT_AUDIO_METHOD);

Also - if we have only one audio subsystem compiled, the AudioOutputMethod
line in speechd.conf should be ignored.

In this case we don't need __SUNPRO_C checking - the macro should define
"oss" as default output method for Solaris as only acceptable.

I could write patch, but I'm not very familiar with autotools and there is
big chance somebody would do it better :)

What do you think?

ethanak
-- 
http://milena.polip.com/ - Pa pa, Ivonko!