Cleanup and use cartopy 0.20 for docs

.travis.yml

@@ -8,7 +8,7 @@ notifications:
   email: false
 
 python:
-  - "3.6"
+  - "3.7"
 
 before_install:
   - |

ci/environment.yml

@@ -5,13 +5,12 @@ name: proplot-dev
 channels:
   - conda-forge
 dependencies:
-  - python>=3.6
+  - python==3.7
   - numpy
   - pandas
   - xarray
-  - matplotlib==3.2.1
-  - cartopy==0.19.0.post1
-  - shapely==1.7.1
+  - matplotlib==3.2.2
+  - cartopy==0.20.2
   - ipykernel
   - pandoc
   - python-build

docs/api.rst

@@ -10,25 +10,24 @@ the docs during a python session.
 
 Please note that proplot removes the associated documentation when functionality
 is deprecated (see :ref:`What's New <whats_new>`). However, proplot adheres to
-`semantic versioning <https://semver.org>`__, which means old code that uses deprecated
-functionality will 1) still work and 2) issue warnings rather than errors until the
-first major release (i.e. version 1.0.0).
+`semantic versioning <https://semver.org>`__, which means old code that uses
+deprecated functionality will still work and issue warnings rather than errors
+until the first major release (version 1.0.0).
 
 .. important::
 
    The color transformation functions like `to_rgba` and `scale_luminance` from
    proplot < 0.10.0 can now be found as methods on the new `~proplot.colors.Color`
-   class. Please see `~proplot.colors.Color` for the relevant documentation. Note that
-   old code that uses commands like ``pplt.to_rgba()`` and ``pplt.scale_luminance()``
-   will still work -- but it will result in a deprecation warning.
+   class. Note that old code that uses commands like ``pplt.to_rgba()`` and
+   ``pplt.scale_luminance()`` will still work (but result in a deprecation warning).
 
 .. important::
 
    The documentation for "wrapper" functions like `standardize_1d` and `cmap_changer`
-   from proplot < 0.8.0 like can now be found under individual `~proplot.axes.PlotAxes`
-   methods like `~proplot.axes.PlotAxes.pcolor`. Please see `~proplot.axes.PlotAxes` for
-   the relevant documentation. Note that calling ``help(ax.method)`` in a python session
-   will show both the proplot documentation and the original matplotlib documentation.
+   from proplot < 0.8.0 can now be found under individual `~proplot.axes.PlotAxes`
+   methods like `~proplot.axes.PlotAxes.plot` and `~proplot.axes.PlotAxes.pcolor`. Note
+   that calling ``help(ax.method)`` in a python session will show both the proplot
+   documentation and the original matplotlib documentation.
 
 Figure class
 ============

docs/environment.yml

@@ -4,8 +4,6 @@
 #   use 3.2.1. Probably lots of basemap holdouts for next ~5 years.
 # * basemap geography is weird with the geos >= 3.9.0 required by cartopy >=
 #   0.19, but important to use 0.19 becuase it fixes padding, so live with it.
-# * cartopy 0.20 makes grid labels look nicer, but it requires python 3.9.
-#   will wait a couple years to catch up to them and keep python 3.6 for now.
 # * pyqt5 is needed by pyplot, RTD server *happens* to already have it
 #   but creating local environment will fail.
 # * jinja >= 3.0 and nbsphinx >= 0.8.2 seem to break default ReST roles when
@@ -16,13 +14,12 @@ name: proplot-dev
 channels:
   - conda-forge
 dependencies:
-  - python>=3.6
+  - python==3.8
   - numpy
   - pandas
   - xarray
-  - matplotlib==3.2.1
-  - cartopy==0.19.0.post1
-  - shapely==1.7.1
+  - matplotlib==3.2.2
+  - cartopy==0.20.2
   - ipykernel
   - pandoc
   - pip

docs/why.rst

@@ -325,18 +325,18 @@ concise figures.
   commands can easily draw colorbars and legends intended to reference more than
   one subplot in arbitrary contiguous rows and columns. See the
   :ref:`next section <why_colorbars_legends>` for details.
-* The `~proplot.axes.Axes.panel_axes` (shorthand `~proplot.axes.Axes.panel`) commands
-  can draw :ref:`thin panels <ug_panels>` along the edges of subplots. This
-  can be useful for plotting 1D summary statistics alongside 2D plots.
 * A-b-c labels can be added to subplots simply using the :rcraw:`abc`
   setting -- for example, ``pplt.rc['abc'] = 'A.'`` or ``axs.format(abc='A.')``.
   This is possible because `~proplot.figure.Figure.add_subplot` assigns a unique
   `~proplot.axes.Axes.number` to every new subplot.
 * The `proplot.gridspec.SubplotGrid.format` command can easily format multiple subplots
   at once or add colorbars, legends, panels, twin axes, or inset axes to multiple
   subplots at once. A `~proplot.gridspec.SubplotGrid` is returned by
-  `proplot.figure.Figure.subplots`, and can be indexed like a list or like a 2D
-  array (in which case the indices match the subplot grid extents).
+  `proplot.figure.Figure.subplots`, and can be indexed like a list or a 2D array.
+* The `~proplot.axes.Axes.panel_axes` (shorthand `~proplot.axes.Axes.panel`) commands
+  draw :ref:`thin panels <ug_panels>` along the edges of subplots. This can be useful
+  for plotting 1D summary statistics alongside 2D plots. You can also add twin axes and
+  panel axes to several subplots at once using `~proplot.gridspec.SubplotGrid` commands.
 
 Links
 -----
@@ -798,19 +798,22 @@ This may be confusing for users outside of the United States.
 Solution
 --------
 
-Proplot introduces the physical units engine `~proplot.utils.units` for
-interpreting `figsize`, `figwidth`, `figheight`, `refwidth`, `refheight`,
-`left`, `right`, `top`, `bottom`, `wspace`, `hspace`, and keyword arguments in a
-few other places. Acceptable units include inches, centimeters, millimeters,
+Proplot uses physical units for the `~proplot.gridspec.GridSpec` keywords
+`left`, `right`, `top`, `bottom`, `wspace`, `hspace`, `pad`, `outerpad`, and
+`innerpad`. The default unit (assumed when a numeric argument is passed) is
+`em-widths <https://en.wikipedia.org/wiki/Em_(typography)>`__. Em-widths are
+particularly appropriate for this context, as plot text can be a useful "ruler"
+when figuring out the amount of space you need. Proplot also permits arbitrary
+string units for these keywords, for the `~proplot.figure.Figure` keywords
+`figsize`, `figwidth`, `figheight`, `refwidth`, and `refheight`, and in a
+few other places. This is powered by the physical units engine `~proplot.utils.units`.
+Acceptable units include inches, centimeters, millimeters,
 pixels, `points <https://en.wikipedia.org/wiki/Point_(typography)>`__, `picas
-<https://en.wikipedia.org/wiki/Pica_(typography)>`__, and `em-heights
-<https://en.wikipedia.org/wiki/Em_(typography)>`__ (a table of acceptable units
-is found :ref:`here <units_table>`). Em-heights are particularly useful, as the
-figure text can be a useful "ruler" when figuring out the amount of space you
-need. The `~proplot.utils.units` function also translates rc settings assigned
-to `~proplot.config.rc_matplotlib` and `~proplot.config.rc_proplot`, e.g.
-:rcraw:`axes.labelpad`, :rcraw:`legend.handlelength`, and
-:rcraw:`subplot.refwidth`.
+<https://en.wikipedia.org/wiki/Pica_(typography)>`__, and (a table of acceptable
+units is found :ref:`here <units_table>`). Note the `~proplot.utils.units` engine
+also translates rc settings assigned to `~proplot.config.rc_matplotlib` and
+`~proplot.config.rc_proplot`, e.g. :rcraw:`subplots.refwidth`,
+# :rcraw:`legend.columnspacing`, and :rcraw:`axes.labelpad`.
 
 Links
 -----

proplot/constructor.py

@@ -1525,7 +1525,7 @@ def Proj(
             raise RuntimeError(
                 'Basemap is no longer maintained and is incompatible with '
                 'matplotlib >= 3.3. Please use cartopy as your geographic '
-                'plotting backend or downgrade to matplotlib <= 3.2.'
+                'plotting backend or downgrade to matplotlib < 3.3.'
             )
         reso = _not_none(
             reso=kwargs.pop('reso', None),