< Back to IRCAM Forum

OMCS 1.3 - Stack Overflow

OpenMusic
#1

Hi there,

Coming to grips with the OMCS Library. I know that it will be an invaluable tool for the way I want to use OM in my pre-compositional phase. However, a slight issue. I seem to frequently and very easily be running into a stack overflow error.

I’ve done some searching and came across this thread which suggested removing some files in the >sources folder for OMCS, but the library has been updated since then and the fix did not work even after trying it. Potential Fix Here

I’ve constructed a fairly simple patch that works up until a certain value is specified. This is all with the :once tag for the but the issue persists (obviously) when requesting 2 or 3 - the idea of :all is completely out of the question!

Additionally, some of the tutorials (n-queens tutorial, pascal triangle) throw a stack overflow error when trying to evaluate the engine using the default patch settings.

Could those of you on mac or linux give this a go and see if the issue persists? If I’m being a complete dingus and this is exactly how the engine works - please set me straight! Any recommended fixes so I can venture forth and make use of the OMCS library?

bad-seq
bad-seq1197×883 40.6 KB

good-seq
good-seq1151×859 33.6 KB

seq-listener
seq-listener1807×307 12.3 KB

GEN Summing Sequence.omp (14.2 KB)

immer dankbar,

Austin

#3

Dear Austin,

Yes you are right here. The problem comes from the compiled sources. Using om7 which is compiled with LW8 we need to recompile the sources (that’s what i did).
Here is the latest version (git) with compiled sources for Linux and Windows:
OMCS.zip (587.5 KB)

This will work.
(mac version coming soon)

Best
K

#4

Dear Austin

A new version (1.4) has been released compiled for all platforms here:

Best
K

2 Likes
#5

What a quick turnaround, thank you Karim!

Upon testing the new 1.4 release on the example patches they work as one would expect. On my provided patch with the above parameters I am able to check for :all solutions which will have the engine run with no stack overflow errors.

Is there an upper limit to the search capacity of the engine? For instance, does it have a set period where it will time-out due to memory usage or size?

I gave it quite a large problem to solve and was given this as the result in the listener window;

User time    =  0:02:47.171
System time  =        4.468
Elapsed time =  0:02:49.020
Allocation   = 14131266664 bytes
0 Page faults
GC time      =       21.640

Can you help me to understand how the “Allocation” works? Is this the total memory used to compute the search?

I appreciate your time and your work! :relieved:

#6

Dear Austin,

Sorry cannot really help you out with this one, since this is the internal memory allocation for the engine’s computing. GC refers to the garbage collecting i believe, so this is some memory allocation exchange for the solver.

However to have an idea how this works, you can take a look at the author’s documentation (p. 43):

It is old but still valid.

Best
K