Nope, I'm not going to suggest a complete threading API here. That -
if it's ever going to be part of XAP - will have to way until we know
what we're doing. (Use pthreads for now. Just don't get any ideas
about toolkits and stuff...)
What I *am* going to suggest is this:
XAP host call:
/*
* Calls a function as a "background job".
* 'context' is the worker call thread ID.
* 'data' is passet to the worker callback
*/
int (*worker_call)(XAP_host *host,
int (*callback)(),
int context, void *data);
XAP event:
/*
* Notify a plugin that one of it's worker
* calls has returned.
*/
XAP_A_WORKER_DONE(int result, void *data)
'context' is used to handle serializing when you want to prevent
multiple workers running at the same time. For example, saying "1"
for all workers you start guarantees that only one of them will run
at a time. If you use different context IDs, the workers may run
concurently on different CPUs, for example.
'data' is user defined data that is passed to the worker thread.
Obviously, you should keep your hands off this data until you get it
back (through XAP_A_WORKER_DONE), since the worker is supposed to be
running in a different thread. You may break this rule if you
*really* know what you're doing. Lock-free FIFOs and similar
constructs that are thread safe by design may be shared. Do note,
however, that hosts are not *required* to actually run worker calls
in another thread!
<maybe>
If this doesn't work for you, you must have a hint
"REQUIRES_WORKER_THREAD" that tells hosts that don't provide
out-of-thread worker calls to stay away from your plugin.
FIXME: Plugins that will work in different ways depending on
whether workers are in a separate thread or not, should
probably be able to tell, preferably during instantiation.
</maybe>
The XAP_A_WORKER_DONE event is sent to the calling plugin as the
worker call returns. 'result' is the return value from the callback,
and 'data' is the 'data' argument passed to the worker call through
host->worker_call().
//David Olofson - Programmer, Composer, Open Source Advocate
.- The Return of Audiality! --------------------------------.
| Free/Open Source Audio Engine for use in Games or Studio. |
| RT and off-line synth. Scripting. Sample accurate timing. |
`---------------------------> http://olofson.net/audiality -'
--- http://olofson.net --- http://www.reologica.se ---
We concluded a while ago that string events are really rather handy
for referencing external files. (For that to be truly useful, hosts
need to be able to understand what these strings are; ie directory
names, file names, temporary storage, output, input. This, so they
can keep track of what belongs in a project.)
Anyway, what I'm seeing here is that strings and raw data blocks
overlap quite significantly on the protocol level. Both are just
control ports that take or send "strings" - either NULL terminated,
or "pascal style". The latter *could* be replace former... Just hint
them differently. Same transfer protocol.
Problem with NULL terminated strings is that they cannot take binary
data (without encoding), and that you have to scan them for the
terminator to find out their length.
OTOH, sending a C string as raw data means you have to do this, even
if the receiver will just ignore the "length" field, and treat it as
the C string it actually is...
I'm leaning towards the "strings in raw data blocks" approach,
despite the little strlen() inconveniency for senders. Mostly because
it's one control data type less to care about, without loss of real
functionality.
Ideas?
//David Olofson - Programmer, Composer, Open Source Advocate
.- The Return of Audiality! --------------------------------.
| Free/Open Source Audio Engine for use in Games or Studio. |
| RT and off-line synth. Scripting. Sample accurate timing. |
`---------------------------> http://olofson.net/audiality -'
--- http://olofson.net --- http://www.reologica.se ---
> Paul Davis <paul(a)linuxaudiosystems.com> writes:
>
> i am also mildly suprised by the value of the deal. steinberg is
> effectively valued at US$24 million.
They sold at a discount to Emagic ... Apple paid 30M for Emagic,
and Apple is not known for over-paying for smaller acquisitions
(NeXT is a different story, although one could argue that Steve
as iCEO was worth every penny ...).
-------------------------------------------------------------------------
John Lazzaro -- Research Specialist -- CS Division -- EECS -- UC Berkeley
lazzaro [at] cs [dot] berkeley [dot] edu www.cs.berkeley.edu/~lazzaro
-------------------------------------------------------------------------
Hi,
I have just bought myself the Cardbus+Multiface RME system. I think I
have finally gotten the alsa modules (snd-hammerfall-mem and snd-hdsp)
to load correctly along with the OSS emulation modules. BUT I cant seem
to play anything out of the Lineout/Headphone output. I have tried
various 'amixer cset' commands but all I get is silence (trying to play
a simple wav file using aplay).
Any suggetsions?
--
---------------------------------------
D. Sen, PhD
21 Woodmont Drive
Randolph
NJ 07869
Home Email: dsen(a)homemail.com Tel: 973 216 2326
Work Email: dsen(a)ieee.org Web: http://www.auditorymodels.org/~dsen
I just had to come out of lurk mode for this:
tim wrote:
> all of them.
>
> rhythmn is always based on one integral periodic 'pulse'. if
> time is not divisible by this atom, there is no musical time.
Nancarow, Ives, Stockhausen, Xenakis, Boulez, Schaeffer, Henry etc. etc in
the classical field
Taylor, Sun Ra, Ornette Coleman, Coltrane, Mengelberg, Broetzman, Zorn,
Ayler etc etc in jazz/impro
lots of ambient stuff that I don't know the names of.
lots of acapella vocal music from various cultures.
There can be easily multiple time-frames going happening in a single piece
of music that have non-lineair relationships.
A computer can also be used to make sounds that a player cannot make.
A sequencer/daw will also be used for non-musical ordering of sounds in
time. It might be handy to use an extended beat/measure structure for
setting event frames for dialog editing for a radio play.
BTW measure are much more complicated than just A/B. Even a 6/8 is really 2/
2.6666666.... in a way. Unless it is divided differently. See Brahms for
nice examples of playing with the groupings of eight notes in 4/4.
The notation x/y is just a shorthand in classical music _notation_, that
only becomes meaningfull in the context of other notation parameters, such
a note-beam groupings etc.
So notating 17/16 instead of 4.25/4 is fine, because the score gives the
grouping information. (to the player and conductor)
Although I have written (4+1/2) / 4 because I wanted to mae sure that the
piece s counted that way and not in 9/8 (=3+3+3).
Anyway my point is that the A/B concept of measure if only really relevant
if your dealing with western _notation_, and then together with the entire
score.
going back to lurk mode now
Gerard
You move the play position marker.
Plugins get the position changes from the timeline,
and those that need to, do their best to prebuffer
audio data from disk, or whatever. While doing that,
the put a "1" on their "READY" Control Outputs, that
are connected to the transport control.
You press "Start".
The transport control simply waits until it has
received a "0" from each one of the "READY" Controls
it's watching. Then it actually starts the sequencer.
If there are no READY Controls in the net, the
sequencer will just start instantly.
Sounds reasonable?
//David Olofson - Programmer, Composer, Open Source Advocate
.- The Return of Audiality! --------------------------------.
| Free/Open Source Audio Engine for use in Games or Studio. |
| RT and off-line synth. Scripting. Sample accurate timing. |
`---------------------------> http://olofson.net/audiality -'
--- http://olofson.net --- http://www.reologica.se ---
The Rosegarden team have the pleasure of announcing the
latest release of their MIDI and audio sequencer and score
editor for Linux.
The source code is available now from the project homepage:
http://www.all-day-breakfast.com/rosegarden
The availability of binary packages depends on their various
maintainers; please check the project homepage for more
information.
New features since the 0.8 release include:
o Improved MIDI file I/O: better support for banks and
merging in import, better support for delay, transpose
etc in export
o MIDI device Bank and Program editor, including import
and export of Studio data
o MIDI Panic Button for clearing down stuck notes
o Added some keyboard controls to matrix
o Added real-time segment delays
o Progress display completely overhauled
o ALSA clients can be added dynamically (you can change
your soft synth configuration while Rosegarden is running)
o MIDI events filter dialog for MIDI THRU and MIDI record
o MIDI/ALSA recording bug fixed (stopping after so many
recorded events)
o Many bug fixes, tweaks and performance improvements
Unfortunately we've had to drop KDE2 support from this release
onwards as it was getting too difficult to maintain both KDE2
and KDE3 in the same development tree. We're hoping this won't
affect too many users in the long term.
Finally, we would like to thank the cachegrind/kcachegrind team
for a truly useful development tool.
Chris
Hi, guys,
I am working on a project in which I need to implement
palyback and recording on the same sound card and in
the same time.
I open the soundcard with RDWR mode and used 'select'
to wait for sound data on card and read it. After
that, I receive audio data from socket and write this
data into buffer on card. for now. I tested it and i
found if just one way, ie. just read or write, it
works fine, the quality of sound is fine. but if two
way, ie. play&record, the sound is horrible and the
weird is that I always get better sound quality from
playback that what i got from recording. (I heard the
audio from two ends). by the way, I used
SNDCTL_DSP_TRIGGER to syncronyse them.
can this method implement full-duplex functions ? if
not, what should I do? thanks in advance
leo
__________________________________________________
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com
I was just thinking about the details of Voice vs Channel Controls.
The events are identical in format and semantics, except that Voice
Controls must have a Virtual Voice ID argument.
As a result, if you have only Channel controls, those could be
exported as Voice controls, if you say you have only 1 Voice. (*)
Seeing it from the sender side, we can also conclude that you can
control a single voice synth with Channel Control events, *provided*
the synth really ignores the VVID. (Which we cannot assume just like
that, of course.)
Or, we could require that Channel Control events carry VVIDs as well!
We wouldn't really have to waste VVIDs on this, because when
connecting Channel Control outputs to Channel Control inputs, no
VVIDs are needed. (They will be ignored, since Channel Controls can't
have that dimension of addressing.)
When connecting a Channel Control output to a Voice Control input, a
fixed VVID will be allocated, giving you a single voice to do what
you want with - as if you were controlling a monophonic synth!
If you're into polyphony, you'll want to be able to grab a bunch of
VVIDs, so you can control multiple Voices. Obiously, this is what
most sequencers will do. But then, what happens if you connect a
sequencer to a monophonic plugin, that has only Channel Controls?
Well, sending Voice Control events with a single (possibly fake) VVID
to Channel Controls works just fine... :-)
We just have to tell the sequencer that there really is only one
Voice. It'll ask for a VVID for that, and the host will go "aha;
Voice -> Channel" and hand the sequencer a fake VVID. (0, that is.)
I'll have to think some more about this, but I think it would be
possible to handle channel->voice and voice->channel Control
connections pretty much transparently. I think this is quite
important, especially for monophonic synths (controlling them from a
standard sequencer), and for building monophonic control processing
nets that run polyphonic synths. (Each Voice will act as a mono
synth.)
(*) Which is probably something we should support, BTW. Some synths
may have a fixed number of voices that are not independent, and
it might be useful to be able to express that in some way.
Documentation or naming might be sufficient, though, as it's
really a matter of how you control the synth. A dual voice
"interference" synth would probably best be controlled from two
sequencer tracks, set to use only one Virtual Voice each - ie
monophonic tracks, like on a traditional tracker.
//David Olofson - Programmer, Composer, Open Source Advocate
.- The Return of Audiality! --------------------------------.
| Free/Open Source Audio Engine for use in Games or Studio. |
| RT and off-line synth. Scripting. Sample accurate timing. |
`---------------------------> http://olofson.net/audiality -'
--- http://olofson.net --- http://www.reologica.se ---
