Andreas Klöckner

Yes, totally unrelated. Loopy CI got broken by https://github.com/inducer/islpy/pull/150 and then broken a bit more by https://github.com/inducer/pymbolic/pull/186. It'll get fixed up by https://github.com/inducer/loopy/pull/937. As long as it's just the same...

This should pass once pyopencl 2025.2 (which I've just released) is on conda-forge. Optimistically tomorrow.

https://github.com/conda-forge/pyopencl-feedstock/pull/130

> it seems like there's a change in behavior.. that -17 became a -16. Yes, there is. > Is that ok? Good question! This had me concerned, too, so I...

Thanks for the report! I can reproduce what you describe, and I can offer a workaround: add `copy_event.wait()` to the end of the routine. This helps because user-initiated `wait` releases...

> I am curious: what is actually the expected behavior here? [islpy](https://github.com/inducer/islpy) wraps [isl](https://libisl.sourceforge.io/), which exposes (among other things) operations on subsets of $\mathbb Z^n$, specifically `BasicSet`s, which are (quasi-)convex,...

Thanks for the contribution! Unfortunately, I don't think this is quite the right approach, and I'm not super sold anything needs fixing. First, it's important to distinguish two scenarios: If...

As you can probably guess, I'm not able to reproduce this. I'm guessing from the font that you might be using pudb on Windows. (Please confirm?) If so, I'm not...

> However, it was only reproducible in some python environment Can you tell me how to make a Python environment in which the problem occurs? Otherwise I feel I won't...

Peculiar pattern in `node_memory_Mapped_bytes`: ([Internal link](http://lager.cs.illinois.edu:9090/graph?g0.expr=node_memory_Mapped_bytes%7Bjob%3D%22relate%22%7D%0A&g0.tab=0&g0.display_mode=lines&g0.show_exemplars=0&g0.range_input=1w))