[CSOUND-DEV:4191] Re: POLL_EVENTS problem solved!

Date	2004-02-28 16:15
From	"gogins@pipeline.com"
Subject	[CSOUND-DEV:4191] Re: POLL_EVENTS problem solved!
	I think it got commented out by mistake, perhaps during a merge. I have uncommented it in my code. After I test it I will commit it. Original Message: ----------------- From: ramsdell@mitre.org (John D. Ramsdell) Date: 28 Feb 2004 10:52:55 -0500 To: csound-dev@eartha.mills.edu Subject: [CSOUND-DEV:4190] Re: POLL_EVENTS problem solved! I deleted my csound5 sandbox and checked out a fresh version of the sources yesterday. I received a version of insert.c with the code to which you refer commented out. Do I have the wrong version? John Engine/insert.c:687:---------------------------------- \|\| (O.OrcEvts && (sensType = sensOrcEvent()))) /* or triginstr event (re Aug 1999) / return(kreq - kcnt); / do early return / } / #if defined(mills_macintosh) \|\| defined(SYMANTEC) / / else if (O.Midiin && actanchor.nxtact == NULL) /\* no midi or notes on; check events \/ / /* #endif / / if (!POLL_EVENTS()) longjmp(cenviron.exitjmp_,1); / kcounter += 1; global_kcounter = kcounter; / IV - Sep 8 2002 / "gogins@pipeline.com" writes: > I'm sorry, but I've never been able to follow diff logs like this. > > However, as best as I can tell, this isn't quite the same as what I had in > mind. kperf is called with N kperiods to compute, in order to do silent > periods as quickly as possible. We both call POLL_EVENTS (or csoundYield, > which had better be exactly the same thing) just above of the silence > block. But I also call it at every kperiod just above the instrument > performance loop, while you seem to have removed that line. > > There probably isn't a whole of difference in practice, but my code might > provide a slightly more responsive GUI. It might be worth trying it both > ways. > > In other words, I have: > > if (!POLL_EVENTS()) longjmp(cenviron.exitjmp_,1); > kcounter += 1; > global_kcounter = kcounter; / IV - Sep 8 2002 / > if (O.sfread) / if audio_infile open / > (spinrecv)(); /* fill the spin buf / > spoutactive = 0; / make spout inactive / > ip = &actanchor; > while ((ip = ip->nxtact) != NULL) { / for each instr active / > pds = (OPDS )ip; > while ((pds = pds->nxtp) != NULL) { > (pds->opadr)(pds); / run each opcode / > } > } > if (spoutactive) / results now in spout? / > (spoutran)(); /* send to audio_out / > else (nzerotran)(1L); /* else send zerospout / > } while (--kcnt); / on Mac/Win, allow system events / > > > > > Original Message: > ----------------- > From: ramsdell@mitre.org (John D. Ramsdell) > Date: 28 Feb 2004 10:05:29 -0500 > To: csound-dev@eartha.mills.edu > Subject: [CSOUND-DEV:4188] Re: POLL_EVENTS problem solved! > > > "Michael Gogins" writes: > > > In Csound5, I have moved the POLL_EVENTS call out of the loop. It is now > > called once per kperiod. > > > > This should be adequate as far as I know, but I would like to hear from > > anyone with reason to think otherwise. > > I think I made the equivalent change to csound4's insert.c. Please > complain if my change is wrong. > > [ramsdell@localhost csound]$ sfcvs csound diff -u > ramsdell@cvs.csound.sourceforge.net's password: > > ? INSTALL > ? Makefile > ? Makefile.in > ? aclocal.m4 > ? autom4te.cache > ? compile > ? config.guess > ? config.log > ? config.status > ? config.sub > ? configure > ? csound.spec > ? depcomp > ? install-sh > ? libtool > ? ltmain.sh > ? missing > ? mkinstalldirs > ? anal/Makefile > ? anal/Makefile.in > ? anal/adsyn/.deps > ? anal/adsyn/Makefile > ? anal/adsyn/Makefile.in > ? anal/adsyn/hetro > ? anal/convol/.deps > ? anal/convol/Makefile > ? anal/convol/Makefile.in > ? anal/convol/cvanal > ? anal/lpc/.deps > ? anal/lpc/Makefile > ? anal/lpc/Makefile.in > ? anal/lpc/lpanal > ? anal/pvoc/.deps > ? anal/pvoc/Makefile > ? anal/pvoc/Makefile.in > ? anal/pvoc/pvanal > ? csound/.deps > ? csound/English.xmg > ? csound/Makefile > ? csound/Makefile.in > ? csound/config.h > ? csound/config.h.in > ? csound/csound > ? csound/csound-config > ? csound/makedb > ? csound/sdif2ad > ? csound/stamp-h1 > ? doc/Makefile > ? doc/Makefile.in > ? doc/csoundgbs.html > ? doc/csoundgbs.info > ? doc/mdate-sh > ? doc/stamp-vti > ? doc/texinfo.tex > ? util1/Makefile > ? util1/Makefile.in > ? util1/scot/.deps > ? util1/scot/Makefile > ? util1/scot/Makefile.in > ? util1/scot/scot > ? util1/sortex/.deps > ? util1/sortex/Makefile > ? util1/sortex/Makefile.in > ? util1/sortex/extract > ? util1/sortex/scsort > ? util2/Makefile > ? util2/Makefile.in > ? util2/dnoise/.deps > ? util2/dnoise/Makefile > ? util2/dnoise/Makefile.in > ? util2/dnoise.dir/srconv > ? util2/envext/.deps > ? util2/envext/Makefile > ? util2/envext/Makefile.in > ? util2/mixer/.deps > ? util2/mixer/Makefile > ? util2/mixer/Makefile.in > ? util2/mixer/extractor > ? util2/mixer/mixer > ? util2/pvlook/.deps > ? util2/pvlook/Makefile > ? util2/pvlook/Makefile.in > ? util2/scale/.deps > ? util2/scale/Makefile > ? util2/scale/Makefile.in > ? util2/sndinfo/.deps > ? util2/sndinfo/Makefile > ? util2/sndinfo/Makefile.in > ? util2/sndinfo/sndinfo > cvs diff: Diffing . > cvs diff: Diffing CsoundCOM > cvs diff: Diffing CsoundVST > cvs diff: Diffing CsoundVST/examples > cvs diff: Diffing CsoundVST/index_files > cvs diff: Diffing CsoundVSTWin > cvs diff: Diffing OSC-Kit > cvs diff: Diffing OSC-Kit/send+dump > cvs diff: Diffing anal > cvs diff: Diffing anal/adsyn > cvs diff: Diffing anal/convol > cvs diff: Diffing anal/lpc > cvs diff: Diffing anal/pvoc > cvs diff: Diffing cscofils > cvs diff: Diffing csound > Index: csound/insert.c > =================================================================== > RCS file: /cvsroot/csound/csound/csound/insert.c,v > retrieving revision 1.4 > diff -u -r1.4 insert.c > --- csound/insert.c 31 Jan 2004 23:10:09 -0000 1.4 > +++ csound/insert.c 28 Feb 2004 14:53:12 -0000 > @@ -647,6 +647,7 @@ > #endif > > VMSG(printf("perfing %ld kprds\n",kcnt);) > + if (!POLL_EVENTS()) longjmp(cglob.exitjmp,1); > if (!O.ksensing && > actanchor.nxtact == NULL) { / if !kreads & !instrs_activ, / > kcounter += kcnt; > @@ -674,10 +675,6 @@ > \|\| (O.OrcEvts && (sensType = sensOrcEvent()))) / or triginstr > event (re Aug 1999) / > return(kreq - kcnt); / do early return / > } > -#if defined(mills_macintosh) \|\| defined(CWIN) \|\| defined(SYMANTEC) > - else if (O.Midiin && actanchor.nxtact == NULL) / no midi or notes > on; check events / > -#endif > - if (!POLL_EVENTS()) longjmp(cglob.exitjmp,1); > kcounter += 1; > global_kcounter = kcounter; / IV - Sep 8 2002 / > if (O.sfread) / if audio_infile open / > @@ -685,7 +682,6 @@ > spoutactive = 0; / make spout inactive / > ip = &actanchor; > while ((ip = ip->nxtact) != NULL) { / for each instr active / > - if (!POLL_EVENTS()) longjmp(cglob.exitjmp,1); / PC GUI needs > attention / > pds = (OPDS )ip; > while ((pds = pds->nxtp) != NULL) { > (pds->opadr)(pds); / run each opcode */ > cvs diff: Diffing doc > cvs diff: Diffing pyrun > cvs diff: Diffing soundfonts > cvs diff: Diffing soundfonts/fluidOpcodes > cvs diff: Diffing util1 > cvs diff: Diffing util1/cscore > cvs diff: Diffing util1/scot > cvs diff: Diffing util1/sortex > cvs diff: Diffing util2 > cvs diff: Diffing util2/dnoise > cvs diff: Diffing util2/dnoise.dir > cvs diff: Diffing util2/envext > cvs diff: Diffing util2/exports > cvs diff: Diffing util2/mixer > cvs diff: Diffing util2/mkgraph > cvs diff: Diffing util2/pvlook > cvs diff: Diffing util2/pvlook.dir > cvs diff: Diffing util2/scale > cvs diff: Diffing util2/scale.dir > cvs diff: Diffing util2/sndinfo > [ramsdell@localhost csound]$ > > > -------------------------------------------------------------------- > mail2web - Check your email from the web at > http://mail2web.com/ . -------------------------------------------------------------------- mail2web - Check your email from the web at http://mail2web.com/ .

Date	2004-02-28 16:22
From	ramsdell@mitre.org (John D. Ramsdell)
Subject	[CSOUND-DEV:4192] Re: POLL_EVENTS problem solved!
	"gogins@pipeline.com" writes: > I think it got commented out by mistake, perhaps during a merge. I have > uncommented it in my code. After I test it I will commit it. Oh, good. I was worried the cause was CVS badness. John