Closed gidden closed 4 years ago
do we make geopandas<0.5.0 a hard install requirement?
I would recommend this for current releases (and master). Then when you can be bothered, add geopandas>0.5.0 and make a new (backwards incompatible) release. I don't use mapping so I don't really mind and would prefer to see efforts focussed elsewhere (but if mapping is a priority for you guys, go for it).
Yes, I am +1 this as well for the moment. The problem with making these requirements is that you start to quickly become incompatible with other packages (that in the future have a ">0.5.0" req).
I think at some point you (we) need to hit pause on development and upgrade both pandas and geopandas to be compatible with new versions. There's already conflicts with things that require pandas>0.24.0 and pyam. (I would also suggest streamlining the release process so they can be made more frequently at the same time, but let's see how things go.)
agreed. I think this is on @danielhuppmann's radar too
yes, it's on my radar - afaik, we are compatible with pandas
0.24.1 (they resolved the issue for combine_first()
which caused unit tests to fail in 0.24.0). But the amount of pandas
warnings on the latest version (replacing labels
with codes
etc.) makes looking at unit test output quite painful.
This is causing more issues with dependencies bumping their version numbers, see #260 for a quickfix - this PR needs to be reverted once we move to geopandas==0.5.0
.
Also, readthedocs still claims to be failing to build on Travis, even though it passes...?
closing this issue because region-plots have been removed from pyam for the time being
It appears region plotting is broken with the latest geopandas release, see attached MPL images (we lose France, among other countries). We should decide:
any thoughts @danielhuppmann or @znicholls