Tests are failing

[Vinc0110]

Pandas 3.0 will be requiring pyarrow:
https://pandas.pydata.org/pandas-docs/version/2.2.0rc0/whatsnew/v2.2.0.html#dedicated-string-data-type-backed-by-arrow-by-default

PyArrow will become a required dependency with pandas 3.0 to accommodate this change.

In #1004 all tests were failing because of various DeprecationWarnings due to this upcoming dependency.

I'm not sure what pandas version is being used in the tests or why this is warning is starting to occur now. Anyway, I've already added pyarrow to pyproject.toml for the test requirements in #1004, which solved the issue, but I thought it would be good to document this change here.

[Vinc0110]

Wow, the tests are still a huge mess:

=========================== short test summary info ============================
  ERROR skrf/calibration/tests/test_calibration.py
  ERROR skrf/calibration/tests/test_calibrationSet.py
  ERROR skrf/calibration/tests/test_deembedding.py
  ERROR skrf/io/tests/test_citi.py
  ERROR skrf/io/tests/test_csv.py
  ERROR skrf/io/tests/test_io.py
  ERROR skrf/io/tests/test_mdif.py
  ERROR skrf/media/tests/test_all_construction.py
  ERROR skrf/media/tests/test_coaxial.py
  ERROR skrf/media/tests/test_cpw.py
  ERROR skrf/media/tests/test_mline.py
  ERROR skrf/tests/test_circuit.py
  ERROR skrf/tests/test_convenience.py
  ERROR skrf/tests/test_frequency.py
  ERROR skrf/tests/test_mathFunctions.py
  ERROR skrf/tests/test_network.py
  ERROR skrf/tests/test_network2.py
  ERROR skrf/tests/test_networkSet.py
  ERROR skrf/tests/test_plotting.py
  ERROR skrf/tests/test_qfactor.py
  ERROR skrf/tests/test_static_data.py
  ERROR skrf/tests/test_tlineFunctions.py
  ERROR skrf/tests/test_util.py
  ERROR skrf/tests/test_vectorfitting.py
  ERROR skrf/vi/tests/test_validators.py
  ERROR skrf/vi/vna/hp/tests/test_8510c.py
  ERROR skrf/vi/vna/keysight/tests/test_fieldfox.py
  ERROR skrf/vi/vna/keysight/tests/test_pna.py
  !!!!!!!!!!!!!!!!!!! Interrupted: 28 errors during collection !!!!!!!!!!!!!!!!!!!
  ============================== 28 errors in 8.99s ==============================
  py39: exit 2 (11.15 seconds) /home/runner/work/scikit-rf/scikit-rf> python -m pytest pid=2004

This does not seem related to my changes. It's more like a general issue.

______________ ERROR collecting skrf/tests/test_vectorfitting.py _______________
  ImportError while importing test module '/home/runner/work/scikit-rf/scikit-rf/skrf/tests/test_vectorfitting.py'.
  Hint: make sure your test modules/packages have valid Python names.
  Traceback:
  /opt/hostedtoolcache/Python/3.9.18/x64/lib/python3.9/importlib/__init__.py:127: in import_module
      return _bootstrap._gcd_import(name[level:], package, level)
  skrf/tests/test_vectorfitting.py:3: in <module>
      import skrf
  skrf/__init__.py:20: in <module>
      from . import taper
  skrf/taper.py:31: in <module>
      from scipy  import linspace
  E   ImportError: cannot import name 'linspace' from 'scipy' (/home/runner/work/scikit-rf/scikit-rf/.tox/py39/lib/python3.9/site-packages/scipy/__init__.py)

@FranzForstmayr: Any ideas?

[Ttl]

It's caused by the new scipy version. linspace should be imported from numpy instead of scipy in taper.py.

[Vinc0110]

That was it, thanks!
I did see that error message, but I could not believe this became an issue from one day to another. How is the scipy version changing? Does it depend on some random decision by the github servers?

[Ttl]

CI installs the newest packages of dependencies unless specified otherwise. It seems that this was changed in the latest scipy release.

[pjg0707]

Seems that scipy had a from numpy import * which was removed after scipy v1.3.3.

What I don't understand is what has changed to make this issue appear now.
Pointing to scipy 1.3.3 is likely to cause many other issues.

[FranzForstmayr]

I don't think it was removed, at least it was able to import linspace from 1.11.4 without error or warning.

Python 3.12.1 (tags/v3.12.1:2305ca5, Dec  7 2023, 22:03:25) [MSC v.1937 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import scipy
>>> scipy.__version__
'1.11.4'
>>> from scipy import linspace
>>>

Python 3.10.12 (main, Nov 20 2023, 15:14:05) [GCC 11.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from scipy import linspace
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ImportError: cannot import name 'linspace' from 'scipy' (/home/franz/.cache/pypoetry/virtualenvs/test-OHDMMxWo-py3.10/lib/python3.10/site-packages/scipy/__init__.py)

There was no deprecation warning in 1.11, that's why we missed to change that earlier.
@jhillairet
The main problem is that a fresh skrf venv fails to import skrf now. We should make a new release asap IMO.

[Vinc0110]

The main problem is that a fresh skrf venv fails to import skrf now. We should make a new release asap IMO.

Is it because of my fix to import linspace from numpy instead of scipy in taper.py? My question is: was it a problem I merged this change so quickly? Do we need to undo it?

[FranzForstmayr]

The main problem is that a fresh skrf venv fails to import skrf now. We should make a new release asap IMO.

Is it because of my fix to import linspace from numpy instead of scipy in taper.py? My question is: was it a problem I merged this change so quickly? Do we need to undo it?

Your fix is perfectly fine, but unreleased. So when installing via pip you'll get the old version which tries to import linspace from scipy.

[Vinc0110]

So then maybe a super quick version 0.30.1 release? We should probably take a bit more time to prepare v1.0.0. @jhillairet ?

[jhillairet]

So then maybe a super quick version 0.30.1 release? We should probably take a bit more time to prepare v1.0.0. @jhillairet ?

Yeah, I think so. Will prepare the next release asap.

[pjg0707]

Setting the scipy version to 1.11.4 instead of 1.12 also worked for me.
Something must have been allowing the linspace to be accessible up to that version and it was changed in the latest release. Good that it was soon identified 👍