Sorry for the late reply. We're working on getting a snapshot out of the xstack.git to
You may want to get your code generation up-to-date then...
I'll be glad to integrate your tests somewhere. FYI, the code doesn't have to be
in a public repo.
Common mortals won't be able to run those tests but the automated system and ourselves
The increasing amount of memory is not normal. I had looked into that a while ago and
leaks. I'll need to revisit the new version. The constant amount of processing may be
the workers trying to aggressively steal, never backing off.
On Jul 10, 2013, at 12:47 PM, Benoit Meister <meister(a)reservoir.com> wrote:
Thanks again for your hospitality during the codathon. It was nice team-working.
Rich is OK sharing our thin runtime layer. So we can add (untuned) tests as generated by
R-Stream to the ocr-execute-api branch, and add the ported tests to the new OCR when it is
I will add a probably incorrect program that exposes another issue with OCR. It looks
like OCR consumes an increasing amount of memory and a constant amount of processing (like
3 full cores on my machine).
Should I send you a tarball ? I don't have write permissions on the repo.