Issue converting Cs6 udo to Cs7 udo

General
#1

Heyy

Not sure if this is a result of the pass-by-reference default behaviour in Csound 7, but I get errors when passing a ftgen to the UDO

giLoad ftgen 1, 0, 0, 1, "path/to/file.wav", 0, 0, 1
giload1len = filelen(gSpath1)

opcode FlooperOverlap(amp:k, rate:k, loopstart:k, loopsize:k, crossfade:k, mytab:i, ptr:k, nvoices:i, cnt:i):(a)
	puts("FlooperOverlap",1)
	print mytab ;will print 1.0
	asig = flooper2:a(amp, rate, loopstart, loopsize, crossfade, mytab)

	if cnt < nvoices-1 then
		asig += FlooperOverlap(amp, rate+random(-0.01, 0.01), loopstart, loopstart+0.1, crossfade, mytab, ptr, nvoices, cnt+1)	
	endif

	xout(asig)
endop

instr FloopTest
	idur = p3
	ifnlen = p4
	ptr:k = line(0, idur, 1) * ifnlen
	print giLoad1 ;will print 1.0
	res:a = FlooperOverlap(1, 1, 0, 0.25, 0.05, giLoad, ptr, 4, 0)
	outs res,  res
endin

schedule(FloopTest, 0, 10, giload1len)

I get this error:

instr 1: giLoad1 = 1.000
FlooperOverlap
UDO FlooperOverlap: mytab = 1.000
Invalid ftable no. 0.000000
INIT ERROR in instr 1 (opcode FlooperOverlap) line 37: function table not found

for some reason the mytab value passed to asig = flooper2:a(amp, rate, loopstart, loopsize, crossfade, mytab) is getting lost.

It’s not an issue in the Cs6 style udo

#2

What build of Csound 7 are you using?

#3

It’s whatever version the latest CsoundQT is running I guess?

image
image578×685 63.5 KB

#4

if the code was like i read it, you missed closing paranthesis after the
list of input argiments, and i see no output arguments. the syntax is:
opcode name(inputs):(outputs)
…
endop

#5

I think the syntax is correct, and it should work fine. I was just chatting with Victor, he thinks there is already an issue in the issue tracker that touches on this bug. I’m just about to rush out the door, so I don’t have a chance to look, maybe you can find it and post this example there?

#6

I think maybe it’s this one?

specifically set value to array not working in new style UDO · Issue #2290 · csound/csound · GitHub

#7

Yep can confirm, that’s the issue

this thing of beauty works:

opcode FlooperOverlap(amp:k, rate:k, loopstart:k, loopsize:k, crossfade:k, ftab:i, ptr:k, nvoices:i, cnt:i):(a)
	my_amp:k = amp*1
	my_rate:k = rate*1
	my_loopstart:k = loopstart*1
	my_loopsize:k = loopsize*1
	my_crossfade:k = crossfade*1
	my_tab:i = ftab*1
	my_ptr:k = ptr*1
	my_nvoices:i = nvoices*1
	my_cnt:i = cnt*1
	asig = flooper2:a(my_amp, my_rate, my_loopstart, my_loopsize, my_crossfade, my_tab)

	if cnt < nvoices-1 then
		asig += FlooperOverlap(amp, rate+random(-0.01, 0.01), loopstart, loopstart+0.1, crossfade, ftab, ptr, nvoices, cnt+1)	
	endif
	xout(asig)
endop
#8

You can also do this instead

opcode Flooper2,a,kkkkkiOoo
amp:k, rate:k, loopstart:k, loopsize:k, crossfade:k, mytab:i, ptr:k, nvoices:k, cnt:i xin
asig = flooper2:a(amp, rate, loopstart, loopsize, crossfade, mytab)
xout(asig)
endop


opcode FlooperOverlap(amp:k, rate:k, loopstart:k, loopsize:k, crossfade:k, mytab:i, ptr:k, nvoices:i, cnt:i):(a)
	puts("FlooperOverlap",1)
	print mytab ;will print 1.0
	asig = Flooper2(amp, rate, loopstart, loopsize, crossfade, mytab)

	if cnt < nvoices-1 then
		asig += FlooperOverlap(amp, rate+random(-0.01, 0.01), loopstart, loopstart+0.1, crossfade, mytab, ptr, nvoices, cnt+1)	
	endif

	xout(asig)
endop

and it will work. The problem seems to be specific to flooper2/flooper - all other opcodes I tried seemed to work.

#9

Thanks @vlz, I got that working. Using the wrapper udo

I did wonder whether something was specifically up with the flooper opcode, as other tests that I’d done didn’t seem to ave trouble with the udo input parameter.

#10

yes, I am still wondering why flooper, but have not found the reason. The original issue was only to do with assignment and the problem is well documented in the PR.

#11

This is now fixed in the develop branch

1 Like