Matplotlib deprecation warnings

[DanShort12]

Hi all,

As well as the rcParams overwrite, mentioned in #460, we see quite a few MatplotlibDeprecationWarning messages when using matplotlib 3.1.2. These stem from lines like below (of which I think there are 5 occurrences):

curr_rc = dict(matplotlib.rcParams)

Where Python will access all of the rcParams in that copy, and so matplotlib will think that any deprecated values are being read. It is possible to copy the rcParams without hitting this warning by calling as below:

curr_rc = matplotlib.rcParams.copy()

That method for getting around this is borrowed from matplotlib/matplotlib#15781.

Apologies, I see the fix for #460 is already in PR #461, I should have probably bundled this observation in with the last issue that I raised.

[GiudGiud]

Hi Dan

There was already a fix for this in develop, with instead of "curr_rc = dict(matplotlib.rcParams)" in plotter.py, there was a slightly modified copy of the dictionary that was already removing some deprecated settings.

Your fix is better however, as it handles automatically the list of deprecated settings, I'm adding it to PR #461

Thanks!

[GiudGiud]

It seems the deprecation warnings still appear because of the line
matplotlib.rcParams.update(curr_rc)
as curr_rc still contains the deprecated parameters (.copy() was silencing warnings, not removing them)

matplotlib/matplotlib#13118 mentions this. The fix implemented catches the deprecation warnings using the warnings library.

[DanShort12]

Ah, I see that the update would also perform a similar access to rcParams, which would generate the warnings too. Catching the warnings when you perform the update seems like a sensible approach to me - that way you'd still get warnings if a deprecated parameter was used elsewhere. Cheers!