If all performances ended normally, it would only be necessary to call reset
after performing. But if a performance is interrupted, I thought it might be
necessary to call reset before performing again. I also found that putting a
call to reset before performing helped me catch some bugs in other reset
code (not setting pointers to null at the very beginning, failing to check
for null pointers in other reset calls, etc.).

It would be simpler to all reset ONLY before performing, but then if Csound
allocates a huge heap of memory during performance, that remains locked up
and unavailable after performance and until the next performance.

So I think calling reset both before and after performing is actually best.
If that's not the order you found in your graph, then the order should be
changed.

============================================
Michael Gogins
gogins at pipeline period com
Irreducible Productions
CsoundVST, an extended version of Csound for programming music and sound
Available at http://sourceforge.net/projects/csound/
============================================


----- Original Message ----- 
From: "stevenyi" 
To: "Csound Developers Discussion List" 
Sent: Sunday, November 23, 2003 1:07 AM
Subject: [CSOUND-DEV:3463] Musmon, cleanup(), csoundCleanup()


> Hi all,
>
> I noticed in the call graph I generated from running csound4, if using
> ccsound.c and csound.c, reset() gets called twice in a run, once at the
> end of musmon(), and again when calling csoundCleanup().
>
> Looking at csound5 cvs, musmon2 calls reset as well, and maybe there's a
> potential for calling reset() twice here as well if csoundCleanup() is
> called.
>
> I think this would happen only if using the csound host API to run, but
> not if using jpff_glue.c.  Don't know if it's a problem to run cleanup()
> twice in a run, but seemed strange so I thought I'd mention it.
>
> Thanks,
> steven
>
>

if performances are interrupted "cleanly" they should call cleanup() -
the setjmp works pretty good this way, ive had very little problems not
cleaning up properly as long as i check the return of every
performBuffer/perormKsmps call, and then call cleanup() in the api.  only
times that there is not a proper
cleanup is a crash.

also i should mention my previous response, i had a typo and in musmon()
the cleanup() is called when !runincomponents - the actual code is:

    if (runincomponents) {
      if (frsturnon != NULL)           /* if something in turnon list  */
        kturnon();                     /*   turnon now                 */
      return 0;                        /* we exit here playevents later*/
    }
    playevents();                      /* play all events in the score */
    return cleanup();

and  it looks like in csound5,  msumon() is used for running in
components(reentrant via api) and musmon2 is used for oldstyle straight
through, which is why musmon2() calls cleanup() and musmon() doesnt?

-m




On Mon, 24 Nov 2003, Michael Gogins wrote:

> If all performances ended normally, it would only be necessary to call reset
> after performing. But if a performance is interrupted, I thought it might be
> necessary to call reset before performing again. I also found that putting a
> call to reset before performing helped me catch some bugs in other reset
> code (not setting pointers to null at the very beginning, failing to check
> for null pointers in other reset calls, etc.).
>
> It would be simpler to all reset ONLY before performing, but then if Csound
> allocates a huge heap of memory during performance, that remains locked up
> and unavailable after performance and until the next performance.
>
> So I think calling reset both before and after performing is actually best.
> If that's not the order you found in your graph, then the order should be
> changed.
>
> ============================================
> Michael Gogins
> gogins at pipeline period com
> Irreducible Productions
> CsoundVST, an extended version of Csound for programming music and sound
> Available at http://sourceforge.net/projects/csound/
> ============================================
>
>
> ----- Original Message -----
> From: "stevenyi" 
> To: "Csound Developers Discussion List" 
> Sent: Sunday, November 23, 2003 1:07 AM
> Subject: [CSOUND-DEV:3463] Musmon, cleanup(), csoundCleanup()
>
>
> > Hi all,
> >
> > I noticed in the call graph I generated from running csound4, if using
> > ccsound.c and csound.c, reset() gets called twice in a run, once at the
> > end of musmon(), and again when calling csoundCleanup().
> >
> > Looking at csound5 cvs, musmon2 calls reset as well, and maybe there's a
> > potential for calling reset() twice here as well if csoundCleanup() is
> > called.
> >
> > I think this would happen only if using the csound host API to run, but
> > not if using jpff_glue.c.  Don't know if it's a problem to run cleanup()
> > twice in a run, but seemed strange so I thought I'd mention it.
> >
> > Thanks,
> > steven
> >
> >
>
>