Set up GitHub actions

[oschusler]

No description provided.

[oschusler]

@elenbaasc this pipeline fails on two things;

1. In the pyproject.toml file you specify that the python version should be >= 3.10. Are you really sure you don't want to support 3.8 and 3.9? They are still in use
2. There is a rounding error on macOS. This should be fixed for the integration tests

[pablolh]

Hi Olaf,

Thanks for the PR, I will fix the integration test and the Python versions

[pablolh]

@oschusler

In the pyproject.toml file you specify that the python version should be >= 3.10.

Concerning this, I just noticed (as I'm building/testing it) that the quantum device connector has python = "^3.11" in its pyproject.toml
What's the reasoning behind this?

Although, OpenSquirrel doesn't have to follow this tight requirement, I think, and I don't see why it wouldn't support 3.8.

[pablolh]

Ah! Numpy doesn't support Python 3.9 apparently. So it'll be 3.9+

[oschusler]

@pablolh the Quantum Device Connector is one of our internal applications and does not run client sided. This means that we can very much pin a specific python version and don't have to support all possible OS' and python versions.

[oschusler]

Ah! Numpy doesn't support Python 3.9 apparently. So it'll be 3.9+

There is no support in Numpy for < 3.10?

[pablolh]

Sorry, I wrote it wrong.
Numpy 1.25+ doesn't support 3.8, but does support 3.9:
see https://endoflife.date/numpy

We require Numpy 1.26 apparently. So alternatively I can accept Numpy 1.24

[oschusler]

Sorry, I wrote it wrong. Numpy 1.25+ doesn't support 3.8, but does support 3.9: see https://endoflife.date/numpy

We require Numpy 1.26 apparently. So alternatively I can accept Numpy 1.24

That is entirely up to you/the project. It only means that the classifiers and matrix in the actions need to be updated

[pablolh]

I got it working!
I had to use multiple Python restricted dependencies in pyproject.toml:
using numpy 1.24 works with Python 3.8, but not Python 3.12
using numpy 1.26 works with 3.12 but not 3.8

I thought that the Numpy package would specify this and that Poetry would deal with this alone, but no, I was getting no clean error and some build failure. Not sure why, anyway, it works now.

Feel free to approve/merge whenever.