Andreas Klöckner

Why not update `to_string`?

Oh OK. Yeah, I agree we should not change the output data type. But we could change the function to return a more complicated string, e.g. `GPU|ACCELERATOR|xyz`. I wouldn't be...

Thanks for the suggestion. I'm not opposed to having type annotations, but I'm unlikely to work on them myself. I'd be happy to take a PR, though.

I'm OK with supporting that. Longer-term, I would also like to support `assignment_to_subst` on ``` tmp[i,0] = 1 tmp[i,1] = 12 ``` generating a conditional, so this seems like a...

That's a limitation of the current implementation. There are not currently plans to address this. I'd be open to considering a PR that adds a better error message. cc @kaushikcfd

We've observed memory leaks in our solver as well: https://gitlab.tiker.net/inducer/pytential/issues/131. We're working on it, but I'd certainly be grateful for help in debugging. Michal mentioned on the pocl list that...

A good start might be to build PyOpenCL with tracing (`./configure.py --cl-trace=1`) and look for `clCreate*`/`clRetain*` and `clRelease*` counts that don't add up to zero.

Thanks for the report! The way this is *supposed* to work is that the loader that's baked into the pyopencl wheel has that search path baked in: https://github.com/inducer/pyopencl/blob/0b3d0ef92497e6838eea300b974f385f94cb5100/scripts/build-wheels.sh#L43-L44 That points...

That's mysterious. Why does it say "missing global symbol" and then return an address for it? And why does this work on other systems?

Could you use some of the same troubleshooting techniques (strace, ldd) to see why that might be happening?