Memory error in SCS about scs HOT 8 CLOSED

What do you mean by memory error? I can't tell what's happening there. Is it a leak, or accessing unallocated memory or something else?

from scs.

I don't really know. I've never seen a problem like this before. Python says there was a double free or corruption, so my guess is that some array is being freed in SCS that shouldn't have been.

The problem only happens when I run nosetests, which evaluates many functions in parallel. I can't replicate it in the python shell running functions serially, so I wasn't able to get a clearer debugging message about the nature of the memory error.

You can replicate/investigate the problem by going to this snapshot

https://www.terminal.com/snapshot/ebc22d6220b84351c0bbe45e0a3428fbbd8815117127d860e948b369ebc335e4

and running the following:

nosetests cvxpy/tests/test_scs.py

from scs.

Looks like it's got something to do with the warm-start vectors that can be passed in. Still working on it.

from scs.

That makes a lot of sense. It also happens when I run

nosetests cvxpy/performance_tests/test_warmstart.py

from scs.

Ok, I've figured it out, it's that I'm free-ing the warm-starts twice basically. I just looked at that code again and there are a couple of weird things that I would like to fix in one go. I'll push a new version tomorrow, it's getting late here...

from scs.

That's awesome you figured it out! No rush, push when you're ready.

On Mon, Jun 8, 2015 at 5:17 PM, bodono [email protected] wrote:

Ok, I've figured it out, it's that I'm free-ing the warm-starts twice
basically. I just looked at that code again and there are a couple of weird
things that I would like to fix in one go. I'll push a new version
tomorrow, it's getting late here...

—
Reply to this email directly or view it on GitHub
#41 (comment).

from scs.

Just pushed a fix.

from scs.

Great! It works for me now. Can you tag a new version and upload it to pypi?

from scs.