Hi,
On Wed, 16 May 2001, Taylor Barratt wrote:
> > Trust me, we do. At least on alsaraw, unixraw, and ossseq; the interfaces
> > are abstracted sufficiently correctly that it _should_ be happening on
> > win32mci as well, at least as far as I can tell. Does it at least print
> > something during startup saying that it is uploading patches? (It should
> > be, if mt32 mode is enabled, i.e. by passing the "-Mmt32" command line
> > option or setting the appropriate value in the config file.)
>
> "-Mmt32?" I used "-mt32" and the console did list a pile of midi patches it
> was "loading" or preparing
I guess you changed it in the config file then. How long does it take to
print those? It _should_ take several seconds, as we're explicitly
delaying the writes there (i.e. much longer than with -Mmt32gm).
> so I'll take it that was a type-o.
No; '-m' will instruct FreeSCI to tell the script that no mouse support is
present. It should complain about the unknown -t, -3, and -2 options,
though...?
Anyway, if the console print that it loaded the patches, this means that
the appropriate calls went to the win32mci driver. I don't know anything
about how it works, but there's one potentially suspicious line here:
midioutput.dwBufferLength = 4; /* size of buffer */
(line 163). Then again, I don't want to jump to conclusions here...
> No noticable
> message on the MT-32 regarding patch uploading.
OK. At least the green LED should be blinking; this sounds as if it's not
even doing that.
> > I guess we should log output from the win32mci driver and compare it to
> > output from other drivers.
> > Then again, there may be timing issues; does the MT-32 in question signal
> > "Exec. buffer overflow"?
>
> Nope, if it does, it doesn't stay on the LCD long enough to read it. I know
> that's an issue with sierra's SCI on fast machines but I haven't had that
> problem.
This problem arises if the explicit delays we're using don't work for some
reason, or if flushing fails (we used to have this problem when using the
libggi driver with ALSA output).
> Maybe FreeSCI is TOO FAST
This is the first time I've heard _that_ ;->
> that I don't even get to see the overflow
> message. *shrug*
That's improbable; IIRC the MT-32 displays the message for at least a
second or so. Anyway, you would be seeing the green "midi message" LED (or
whatever it's called) blinking during startup (while the messages are
print) if anything was written to it...
llap,
Christoph
