Jed Brown

I am going to make an epistemic argument regarding copyright infringement. First, note that LLMs prompted with the first line of a book [can generate](https://arxiv.org/abs/2505.12546) the entire book verbatim. The...

> We're not a journal and people aren't submitting papers but code. Scholarly legitimacy is necessary, and even more important for the JOSS partnership, which we want universities to "count"...

> If it is a problem, with so much code written with LLMs, there should be many, prominent cases. This prominent class-action case for code was filed in 2022 and...

I think proposing it in Slack would also connect with an audience beyond this issue, and help prime a discussion at the next monthly meeting.

> Are there any other flags of mpiexec that you would like to support besides -n/-np? Process affinity (`-bind-to`, `-map-by`) and GPU management would be a luxury, though not needed...

Oh, I don't want to run jobs sequentially. Rather, use `make -j32 test`, which attempts to always have 32 jobs submitted. Some of those jobs are single-core while others use...

The job slots protocol doesn't give a way to avoid deadlock when running multi-process jobs (which would need to acquire multiple slots to start) -- prior conversation: https://github.com/flux-framework/flux-core/issues/2915#issuecomment-621338594 https://github.com/flux-framework/flux-core/issues/2915#issuecomment-621344325 The...

Cool, that's entering usable territory for our use case. Open MPI `mpiexec` has an overhead of about 30 ms, so it's already getting close. How expensive are the serialized KVS...

Hi, just checking if this is complete (if so, how should I use to observe that fast submission time?) or unplanned.

Would you consider Rust for the submission tool? That would give good library support for CLI handling and JSON construction, and you get a statically-linked executable that's every bit as...