<div dir="ltr">Miller<div><br></div><div>Does that still prevent multiple instances of a libpd based VST from loading in a host?</div><div><br></div><div>Chris</div></div><div class="gmail_extra"><br><br><div class="gmail_quote">
On Mon, May 12, 2014 at 10:12 AM, Miller Puckette <span dir="ltr"><<a href="mailto:msp@ucsd.edu" target="_blank">msp@ucsd.edu</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
I think the global shared symbol space will cause some confusion (patches<br>
will have to protect their sends and receives, etc., by fabricating local<br>
names using '$' characters.<br>
<br>
Also, it won't be possible to run several Pds concurrently on several<br>
threads - in a threaded environment, each Pd instance will have to lock<br>
all others out while it runs. (There might be a fix for that by replacing<br>
all static variables in Pd by thread-local ones, but this is untested and<br>
I'd hate to do it unless there's a real need for it, since it would involve<br>
systematically changing 100s of variable definitions in Pd.)<br>
<br>
I'm not sure about timing - that depends partly on the libpd release process<br>
that I'm not involved in.<br>
<br>
cheers<br>
Miller<br>
<div><div class="h5">On Mon, May 12, 2014 at 09:45:25AM -0400, Rob Bairos wrote:<br>
> Hi Miller.<br>
> Thats very exciting news!<br>
> Can you give a brief synopsis of what you think the biggest remaining<br>
> limitations are and a rough timeline as to when a stable multi-instance<br>
> pdlib would be available for general use?<br>
><br>
> Thanks again!<br>
> Rob Bairos.<br>
><br>
><br>
><br>
> On Sun, May 11, 2014 at 3:00 AM, Miller Puckette <<a href="mailto:msp@ucsd.edu">msp@ucsd.edu</a>> wrote:<br>
><br>
> > To Pd developers...<br>
> ><br>
> > I've adapted Pd to nternally use a "pdinstance" structure to allow<br>
> > multiple schedulers (including DSP chains) to run in one address space.<br>
> > With slight modification (see below) libpd can use this to run separate<br>
> > patches asynchronously.<br>
> ><br>
> > I tried to "instance-ize" all Pd symbols to keep them private but ran into<br>
> > seemingly insoluble problems, so am leaving all symbols global for now.<br>
> > (So patches used in pdlib, if they want to be usable in a multiple-instance<br>
> > scenaro, should protect all send/receive/table/delay/etc named objects with<br>
> > a $[0-9] somewhere.<br>
> ><br>
> > I didn't look hard, but I thnk pdlib now will need a way to send $<br>
> > arguments<br>
> > to patches via libpd_openfile() somehow. Also, libpd_init_audio() will<br>
> > want to<br>
> > make number of channels per-instance, etc. I've put some relevant comments<br>
> > in the test program I'll include below (and sorry for the length of this<br>
> > mail!)<br>
> ><br>
> > Here's how I modified libpd_wrapper/z_libpd.c:<br>
> ><br>
> > 55d54<br>
> > < sys_time = 0;<br>
> > 110c109<br>
> > < sched_tick(sys_time + sys_time_per_dsp_tick);<br>
> > ---<br>
> > > sched_tick();<br>
> > 130c129<br>
> > < sched_tick(sys_time + sys_time_per_dsp_tick); \<br>
> > ---<br>
> > > sched_tick(); \<br>
> ><br>
> > ----------------------<br>
> > and here (sorry again) is the test program, "pdtest2.c" adapted from<br>
> > libpd-master/samples/c_samples/c/pdtest.c :<br>
> > ---------------------<br>
> ><br>
> > #include <stdio.h><br>
> > #include "z_libpd.h"<br>
> > #include "m_imp.h"<br>
> > #define TIMEUNITPERMSEC (32. * 441.)<br>
> ><br>
> > void pdprint(const char *s) {<br>
> > printf("%s", s);<br>
> > }<br>
> ><br>
> > void pdnoteon(int ch, int pitch, int vel) {<br>
> > printf("noteon: %d %d %d\n", ch, pitch, vel);<br>
> > }<br>
> ><br>
> > int main(int argc, char **argv) {<br>
> > t_pdinstance *pd1 = pdinstance_new(), *pd2 = pdinstance_new();<br>
> > if (argc < 3) {<br>
> > fprintf(stderr, "usage: %s file folder\n", argv[0]);<br>
> > return -1;<br>
> > }<br>
> ><br>
> > int srate = 44100;<br>
> > // maybe these two calls should be available per-instnace somehow:<br>
> > libpd_set_printhook(pdprint);<br>
> > libpd_set_noteonhook(pdnoteon);<br>
> > /* set a "current" instance before libpd_init() or else Pd will make<br>
> > an unnecessary third "default" instance. */<br>
> > pd_setinstance(pd1);<br>
> > libpd_init();<br>
> > /* ... here we'd sure like to be able to have number of channels be<br>
> > per-nstance. The sample rate is still global within Pd but we might<br>
> > also consider relaxing that restrction. */<br>
> > libpd_init_audio(1, 2, srate);<br>
> ><br>
> > float inbuf[64], outbuf[128]; // one input channel, two output channels<br>
> > // block size 64, one tick per buffer<br>
> ><br>
> > pd_setinstance(pd1); // talk to first pd instance<br>
> ><br>
> > // compute audio [; pd dsp 1(<br>
> > libpd_start_message(1); // one entry in list<br>
> > libpd_add_float(1.0f);<br>
> > libpd_finish_message("pd", "dsp");<br>
> ><br>
> > // open patch [; pd open file folder(<br>
> > libpd_openfile(argv[1], argv[2]);<br>
> ><br>
> > pd_setinstance(pd2);<br>
> ><br>
> > // compute audio [; pd dsp 1(<br>
> > libpd_start_message(1); // one entry in list<br>
> > libpd_add_float(1.0f);<br>
> > libpd_finish_message("pd", "dsp");<br>
> ><br>
> > // open patch [; pd open file folder(<br>
> > libpd_openfile(argv[1], argv[2]);<br>
> ><br>
> > /* the follownig two messages can be sent without setting the pd<br>
> > nstance<br>
> > and anyhow the symbols are global so they may affect multiple<br>
> > instances.<br>
> > However, if the messages change anyhing in the pd instacne structure<br>
> > (DSP state; current time; list of all canvases n our instance) those<br>
> > changes will apply to the current Pd nstance, so the earlier messages,<br>
> > for instance, were sensitive to which was the current one.<br>
> ><br>
> > Note also that I'm using the fact taht $0 is set to 1003, 1004, ...<br>
> > as patches are opened -it would be better to opent the patches with<br>
> > settable $1, etc parameters to libpd_openfile(). */<br>
> ><br>
> > // [; pd frequency 1 (<br>
> > libpd_start_message(1); // one entry in list<br>
> > libpd_add_float(1.0f);<br>
> > libpd_finish_message("1003-frequency", "float");<br>
> ><br>
> > // [; pd frequency 1 (<br>
> > libpd_start_message(1); // one entry in list<br>
> > libpd_add_float(2.0f);<br>
> > libpd_finish_message("1004-frequency", "float");<br>
> ><br>
> > // now run pd for ten seconds (logical time)<br>
> > int i, j;<br>
> > for (i = 0; i < 3; i++) {<br>
> > // fill inbuf here<br>
> > pd_setinstance(pd1);<br>
> > libpd_process_float(1, inbuf, outbuf);<br>
> > if (i < 2)<br>
> > {<br>
> > for (j = 0; j < 8; j++)<br>
> > printf("%f ", outbuf[j]);<br>
> > printf("\n");<br>
> > }<br>
> > pd_setinstance(pd2);<br>
> > libpd_process_float(1, inbuf, outbuf);<br>
> > if (i < 2)<br>
> > {<br>
> > for (j = 0; j < 8; j++)<br>
> > printf("%f ", outbuf[j]);<br>
> > printf("\n");<br>
> > }<br>
> > }<br>
> ><br>
> > return 0;<br>
> > }<br>
> ><br>
> > ----------------------<br>
> > I replaced "test.c" as follows:<br>
> > ---------------------<br>
> ><br>
> > #N canvas 406 290 450 300 10;<br>
> > #X obj 97 64 loadbang;<br>
> > #X obj 97 131 print;<br>
> > #X obj 186 106 dac~;<br>
> > #X obj 97 107 f 0;<br>
> > #X obj 128 107 + 1;<br>
> > #X obj 97 86 metro 2;<br>
> > #X obj 185 41 r \$0-frequency;<br>
> > #X obj 188 73 osc~;<br>
> > #X obj 248 127 print \$0-frequency;<br>
> > #X obj 248 97 loadbang;<br>
> > #X connect 0 0 5 0;<br>
> > #X connect 3 0 1 0;<br>
> > #X connect 3 0 4 0;<br>
> > #X connect 4 0 3 1;<br>
> > #X connect 5 0 3 0;<br>
> > #X connect 6 0 7 0;<br>
> > #X connect 6 0 8 0;<br>
> > #X connect 7 0 2 0;<br>
> > #X connect 7 0 2 1;<br>
> > #X connect 9 0 8 0;<br>
> ><br>
> > ----------------------------<br>
> > and got this output:<br>
> ><br>
> > print: 0<br>
> > 1003-frequency: bang<br>
> > print: 0<br>
> > 1004-frequency: bang<br>
> > 1003-frequency: 1<br>
> > 1004-frequency: 2<br>
> > 1.000000 1.000000 0.999999 0.999999 0.999998 0.999998 0.999997 0.999997<br>
> > 1.000000 1.000000 0.999998 0.999998 0.999996 0.999996 0.999995 0.999995<br>
> > print: 1<br>
> > 0.999944 0.999944 0.999943 0.999943 0.999942 0.999942 0.999941 0.999941<br>
> > print: 1<br>
> > 0.999815 0.999815 0.999810 0.999810 0.999804 0.999804 0.999799 0.999799<br>
> > print: 2<br>
> > print: 2<br>
> ><br>
> > ... Looks like there are 2 schedulers adn DSP chains running :)<br>
> ><br>
> > Miller<br>
> ><br>
> ><br>
> ><br>
> > _______________________________________________<br>
> > Pd-dev mailing list<br>
> > <a href="mailto:Pd-dev@iem.at">Pd-dev@iem.at</a><br>
> > <a href="http://lists.puredata.info/listinfo/pd-dev" target="_blank">http://lists.puredata.info/listinfo/pd-dev</a><br>
> ><br>
<br>
</div></div>_______________________________________________<br>
Pd-dev mailing list<br>
<a href="mailto:Pd-dev@lists.iem.at">Pd-dev@lists.iem.at</a><br>
<a href="http://lists.puredata.info/listinfo/pd-dev" target="_blank">http://lists.puredata.info/listinfo/pd-dev</a><br>
</blockquote></div><br></div>