numpy 2 breaks on import np.infty

[cstarkjp]

Just a heads-up: NumPy 2 doesn't have infty so queueing_tool.queues.queue_servers breaks.

To get your code working in a conda env, I accidentally upgraded NumPy to 2.0.0 while downgrading Python to 3.10, before doing pip install of version 1.2.5.

Downgrading NumPy back to 1.26.4 fixed the issue.

I got an import error in from queueing_tool.queues import * because queueing_tool.queues.queue_servers tries at some level to do from numpy import infty.

No surprise that np 2 is breaking stuff.

By the way, it would be nice if: (1) there were a conda-forge version available; (2) there were versions available for 3.11, 3.12.

[djordon]

Whoa, I completely missed that they were working on NumPy 2. It appears that the fix is simple, all you need is to us np.inf instead of np.infty.

1. I don't know how to release conda-forge versions. The PyPI version doesn't work with conda-forge (am I even saying this right?)?
2. It should be easy release versions that work with python 3.11 and 3.12.

[cstarkjp]

Good to know it's an easy fix. I would generate a PR, but it sounds like it's not really necessary in this case.

Context here: I'm preparing some code to accompany an academic paper (in geoscience). To make it easier for my non-techie colleagues, I try to make installation as simple and as bundled as possible, so I use conda and its environment.yml. So far queueing_tool is the only package that I have to pip install and break the consistent environment in the process.

Conda-forge is (Ana)conda's most open distribution set. I don't have practical experience of getting a package in there, but I bet it's not much more hassle than for PyPi. If you can figure it out, it will expand the adoption of your code!