Update docstring in __setitem__

[rohanbabbar04]

• Update docstring in __setitem__

[rohanbabbar04]

@mrava87, Can you tell me where are the tests failing, as you have mentioned it here? I can check it at my end.

[mrava87]

@rohanbabbar04 i didn't mean actual unittests but more tests I was doing... if you take this https://github.com/PyLops/pylops-nccl_examples/blob/main/bench_mdd.py and replace the UNSAFE_BROADCAST you will see that things start to hung, I didn't have time to investigate why but I'm quite sure it is something to do with the comm we do in setitem as skipping it (when UNSAFE_BROADCAST is used) seems to solve the problem

[rohanbabbar04]

Thanks @mrava87 , I will take a look.
Looks like I do not have access to the above mentioned repo.

[mrava87]

Looks like we got numpy2.3.5 installed in this action https://github.com/PyLops/pylops-mpi/actions/runs/22067038064/job/63761469742?pr=185

[mrava87]

As suspected, I retrigged the CI and everything breaks because now numpy2.4.2 gets installed

https://github.com/PyLops/pylops-mpi/actions/runs/22067038064/job/66000565531?pr=185

Since we saw this when we were just about to release a new version #187, I suggest the following course of action:

• Make new PR forcing numpy<=2.3.5 for now
• Merge release PR and make new release
• Start addressing those numpy issues and relax version again
• Back to all other PRs

@rohanbabbar04 @tharittk ?

PS: completely unrelated... @rohanbabbar04 in the future can you please avoid making branches in the main repo and work off your fork... once we sort these things I also suggest we move to PyLops' strategy of having dev as our main branch where we merge PRs and main as a stable branch we merge into only just before a release (and keep releasing from main)

[mrava87]

@rohanbabbar04 let's merge this if you agree - and you can keep looking into the issue of multi-node multi-gpu setup failing with BROADCAST partition...

[rohanbabbar04]

@rohanbabbar04 let's merge this if you agree - and you can keep looking into the issue of multi-node multi-gpu setup failing with BROADCAST partition...

Sounds good