You follow the Steven's suggestions and, if you want to play with your
example, remember that the `tab' opcode doesn't check the index validity.
You have to use an index between 0 and `inumparts - 1' and not between
1 and `inumparts' (a rapid fix: `loop_lt' instead of `loop_le').

tito

On Sun, Feb 12, 2012 at 06:08:56PM +0100, Stefan Thomas wrote:
> Dear Steven,
> thanks for sending Your opcode to me.
> I will try it, for sure, but first I will study Your articles.
> To be honest, I've started reading it and then I thought "let's do it by
> myself", but that seems to be a bit too early.
> 
> Am 12. Februar 2012 18:00 schrieb Steven Yi <stevenyi@gmail.com>:
> 
> > Hi Stefan,
> >
> > There's a couple of issues here.  First thing is that you're doing an
> > init-time loop though you have performance-time code within it, which
> > should be avoided.  Using k-vars will ensure you are doing the loop at
> > performance time.  Also, you can not loop and call the same opcode
> > over and over when it has its own state.  Your goal is to run multiple
> > modes, but instead you are running the same mode opcode over and over.
> >
> > I'd recommend reviewing two articles of mine regarding control flow:
> >
> > http://www.csounds.com/journal/2006spring/controlFlow.html
> > http://www.csounds.com/journal/2006summer/controlFlow_part2.html
> >
> > particularly part 2 which discusses recursion.  For this scenario, I
> > would recommend using recursion with a UDO if you are going to have
> > n-number of performance time modes.
> >
> > I was working on a recursive mode instrument and made a UDO a couple
> > of months ago.  The instrument is not complete, but the code I used
> > for the UDO is here:
> >
> >        opcode mode2,a,akkkko
> >
> > ain, kfreq, kq, kampmultiplier, kqmultiplier, icounter xin
> >
> > kcurrentfreq = kfreq * (icounter + 1)
> >
> > itie tival
> >
> > aout mode ain, kcurrentfreq, kq, itie
> > ;printk 1, kcurrentfreq
> > if (icounter < 64) then
> >
> > kfreqnext = kcurrentfreq + kfreq
> >
> > if (kfreqnext < (sr * .2) && icounter < 64) then
> >
> > aout2 mode2 ain * kampmultiplier, kfreq, kq * kqmultiplier,
> > kampmultiplier, kqmultiplier, icounter + 1
> > aout sum aout, aout2
> >
> > endif
> >
> > endif
> >
> > xout aout
> >
> >        endop
> >
> > (This UDO is also still a work in progress).
> >
> > Hope that helps and good luck!
> > steven
> >
> >
> > On Sun, Feb 12, 2012 at 4:46 PM, Stefan Thomas
> > <kontrapunktstefan@googlemail.com> wrote:
> > > Dear community,
> > > I'm trying to build my own udo, using the mode-opcode.
> > > My aim:
> > > reading the relative partial-frequencies from a table, as many as I
> > desire,
> > > and stop reading, when the end of the table is reached.
> > > First I've created a csd-file, as a template for the udo. But
> > unfortunately
> > > the code doesn't work.
> > > When I play the first note, I get:
> > > PERF ERROR: tab_i off end
> > > and I can't hear a sound.
> > > Sorry, it's my first trial with iterative code, so maybee I'm a little
> > > thumb.
> > > Here is my code:
> > >
> > >
> > > <CsoundSynthesizer>
> > > <CsOptions>
> > > -odevaudio -Ma -b-1 -B 4096 -m0d
> > > </CsOptions>
> > > <CsInstruments>
> > >
> > > sr    =    48000
> > > ksmps     =     100
> > > nchnls    =    2
> > > 0dbfs    =    1
> > >
> > > giPartfreqs    ftgen        0, 0, 8, -2, 1, 2,3,4,5,6,7,8
> > >
> > > instr 1
> > > iamp ampmidi 1
> > > icps cpsmidi
> > > iq = 200
> > > ain mpulse iamp, 0
> > > indx init 0
> > >         inumparts = ftlen (giPartfreqs)
> > > loop:
> > > indx = indx+1
> > > ipartfreq    tab_i        indx, giPartfreqs
> > >
> > > ipartfreq    =        ipartfreq * icps
> > > asig         mode    ain, ipartfreq, iq
> > >     amix    init    0
> > > amix        =  asig+amix
> > > amix         = amix/inumparts
> > >         loop_le    indx, 1, inumparts, loop
> > >
> > > outs amix, amix
> > >
> > > endin
> > >
> > > </CsInstruments>
> > > ; ==============================================
> > > <CsScore>
> > > f 0 36
> > >
> > >
> > > </CsScore>
> > > </CsoundSynthesizer>
> > >
> > >