Andreas Klöckner

This also seems to affect OIDC, in my case using a Microsoft tenant as an IdP. Here's the relevant section of my `docker-compose.yml`: ``` grist: image: docker.io/gristlabs/grist environment: GRIST_DEFAULT_EMAIL: "[email protected]"...

> Do you have `llvm-spirv-19` installed ? I don't. (From reading your comment, that may be a good thing.) I agree that that packaging approach seems... questionable. Btw, in case...

Thanks for the report! a4c265558f7915a250ca222d6b39738ddc9831d0 did some of that. I've just updated the online PDF to reflect that. Does that address what you were looking for, or is anything else...

> Is `unbind_queue` now called [`unbind_from_queue`](https://documen.tician.de/pyopencl/runtime_memory.html#pyopencl.SVMAllocation.unbind_from_queue)? Yes. > Also would you be able to clarify this question a little bit more? I'm struggling to understand what exactly the issue is...

> So the queues that will need to be waited on before deallocation would be the ones that still have pending SVM tasks to do There's no practical way to...

Ooh interesting. Thanks for pointing that out. I might play around some over the summer if/when I have some time. Thanks to `uv`, it's not hard to do: ``` uvx...

The only sense I can make of this traceback is that `Debugger` instance creation must have failed in `__init__`, which means there must be an error message prior to what...

Thanks for the report! #685 should take care of the "outer" exception. Probably the mock-based testing should be ripped out and replaced, too, but I will leave that for another...

Unsubscribing... @-mention or request review once it's ready for a look or needs attention.

> Downstreams seem very upset with it though :( They sure are. I'm trying to figure out whether the damage can be controlled or whether we need to do this...