Handle Pandas time indices within StationPlots #1130
Conversation
zbruick
added
Type: Bug
zbruick
force-pushed
the
stationplot_pandas
branch
from
c2fe4c8
to
875f115
Compare
|
Got the unit handling figured out I think. I added an image test (based on #957), but since this is my first to do, please let me know if I need to change anything. I dropped all of the tolerances down to 0 and it passed on my machine, so I'm curious how CI does with it, or if the test isn't doing what I think it is. |
zbruick
force-pushed
the
stationplot_pandas
branch
3 times, most recently
from
3a7bc05
to
a43fa2b
Compare
|
Do we need to add the Pandas dependency here if we're not actually using it? |
|
Oh, it's used in testing... |
|
We already have it as a testing dependency I believe (#1034). So I think it could be removed from this PR. |
|
I might be missing something, but isn't pandas already a dependency of MetPy, since pandas is a hard dependency of xarray? It would just be determining which versions to support/test? |
|
@zbruick You might cherry-pick the commits onto another of your PRs to save the work. We might ship 0.11 without the other PRs (who knows) so to me it seems better to only add it as a strict dependency when we actually have it. @jthielen While the environment will have Pandas because of Xarray, we should only be listing dependencies on things that MetPy actually imports in the library proper. That way if for some reason xarray drops its dependence on Pandas, we're not forcing it needlessly in our packaging. Probably not a practical difference here, but a good policy to follow. Likewise, we should have a listing for everything we import and not just rely on it being pulled in transitively from something else we depend on. |
|
Yeah I agree with the comments here. No reason to add it in this PR - I forgot I was dropping what I was using it for with the update. Good suggestion to cherrypick them over before dropping them. Fine with me to ship 0.11 without the DSG PR, as I'm guessing the appending stuff I added may take some time to review and revise. |
Resolves issues with Pandas DatetimeIndex.
zbruick
force-pushed
the
stationplot_pandas
branch
2 times, most recently
from
6e1ab33
to
116d522
Compare
|
Pandas dependency has been removed, and this PR has been rebased. |
zbruick
force-pushed
the
stationplot_pandas
branch
from
116d522
to
b54a276
Compare
Description Of Changes
Adds support for Pandas DatetimeIndex and TimedeltaIndex as valid
xvariables inStationPlot. This addresses #957, which shows a use case for a meteogram of sky cover (and weather symbols could be imagined too). If needed, other datetime types could be added for further support.This also adds a dependency on Pandas, which I think we're now up to 3 PRs in the 0.11 queue with this dependency.
Checklist