Plotly

Updated
- Updated Plotly.js to version 1.43.1. See the
[plotly.js CHANGELOG](https://github.com/plotly/plotly.js/blob/master/CHANGELOG.md#1431----2018-12-21)
for more information.

Changed
- Plotly.js 1.43 converted `title` properties (e.g. `layout.title`) from
strings into compound objects that contain the text as the `text` property
along with new title placement attributes `x`, `y`, `xref`, `yref`, `xanchor`,
`yanchor` and `pad`. Plotly.py 3.5.0 follows the new schema, but still
supports specifying `title` as a string, in which case the string is assigned
to the `title.text` property
([1302](https://github.com/plotly/plotly.py/issues/1302))
- Plotly.js 1.43 also moved existing `title*` properties
(e.g. `layout.titlefont`) under the `title` object (e.g. `layout.title.font`).
Plotly.py 3.5.0 follows the new schema, but still
supports the legacy `title*` properties by mapping them to the corresponding
`title.*` property
([1302](https://github.com/plotly/plotly.py/issues/1302))
- The `update` method on `graph_objs` now returns the updated object in order
to support chaining multiple update operations together
([1379](https://github.com/plotly/plotly.py/issues/1379))
- The `show_link` option has been set to `False` by default in the offline
`plot` and `iplot` functions. Now that the "send data to cloud" button has
been disabled by default in plotly.js 1.43.0, no buttons/links will be
displayed by default that result in data being sent off of the local machine
([1304](https://github.com/plotly/plotly.py/issues/1304))
- `config` options that are not known by plotly.py result in a warning but are
still passed along to plotly.js. Prior to this change these unknown options
were dropped silently
([1290](https://github.com/plotly/plotly.py/issues/1290))
- Built-in themes now specify colorscales using the new global
`layout.colorscale` properties. Previously the colorscales were defined for
each trace type individually. This reduces the size of the resulting theme
files
([1303](https://github.com/plotly/plotly.py/issues/1303))
- Increased the maximum retry time of the orca integration from 8s to 30s
([1297](https://github.com/plotly/plotly.py/issues/1297))

Fixed
- Fixed `FigureWidget` performance regression that, when working with
large datasets, resulted in a slight freeze of the widget after user
interactions (pan, zoom, etc)
([1305](https://github.com/plotly/plotly.py/issues/1305))
- Fix orca error when the `ELECTRON_RUN_AS_NODE` environment variable is set
([1293](https://github.com/plotly/plotly.py/issues/1293))
- The `'responsive'` config key was being silently blocked
([1290](https://github.com/plotly/plotly.py/issues/1290))
- Fixed error when using unicode characters in string properties on Python 2
([1289](https://github.com/plotly/plotly.py/issues/1289))
- Removed invalid calls to non-existent `validate` and `strip_style` `Figure`
methods in matplotlylib conversion logic
([1128](https://github.com/plotly/plotly.py/issues/1128))

Fixed
- `config` options are now supported when using `plotly.offline.iplot` to
display a figure in JupyterLab. Requires version 0.18.1 of the
`jupyterlab/plotly-extension` extension.
([1281](https://github.com/plotly/plotly.py/pull/1281),
[jupyterlab/jupyter-renderers168](https://github.com/jupyterlab/jupyter-renderers/pull/168))
- Custom `plotly_domain` values are now supported in FigureWidget in both
the classic notebook and JupyterLab
([1284](https://github.com/plotly/plotly.py/pull/1284))

Updated
- Updated Plotly.js to version 1.42.5. See the
[plotly.js CHANGELOG](https://github.com/plotly/plotly.js/blob/master/CHANGELOG.md#1423----2018-11-06)
for more information.

Fixed
- Fixed histogram binning with pandas `Series` or numpy array
(regression introduced in 3.4.0)
([1257](https://github.com/plotly/plotly.py/issues/1257),
[plotly/plotly.js3211](https://github.com/plotly/plotly.js/pull/3211))
- Fixed incorrect validation error on the `args` property of
`layout.updatemenu.Button()` when value is a `list` that starts with a `list`
([1265](https://github.com/plotly/plotly.py/issues/1265))
- Fixed deadlock causing `plotly.io.write_image` to hang on Windows after
exporting more than ~25 images
([1255](https://github.com/plotly/plotly.py/issues/1255))
- Fixed plot display error for `scattergl` trace with `mode='lines'` and
more than 100k points
([1271](https://github.com/plotly/plotly.py/issues/1271))
- Fixed responsive resizing error with `iplot` in the classic notebook
([1263](https://github.com/plotly/plotly.py/pull/1263))

Updated
- Updated Plotly.js to version 1.42.2. Select highlights included below, see
the [plotly.js CHANGELOG](https://github.com/plotly/plotly.js/blob/master/CHANGELOG.md#1422----2018-11-01)
for more information.

Added
- Default figure properties may now be customized using figure
templates (themes) and 7 new predefined templates are bundled with
plotly.py
([1224](https://github.com/plotly/plotly.py/pull/1224))
- Added Parallel Categories (`parcats`) trace type for the visualization
of multi-dimensional categorical datasets
([plotly/plotly.js2963](https://github.com/plotly/plotly.js/pull/2963))
- Added LaTeX typesetting support for figures displayed in the Jupyter
Notebook using `plotly.offline.iplot` and `plotly.graph_objs.FigureWidget`.
**Note:** There are still outstanding issues with MathJax rendering in FireFox,
but it is now working well in Chrome.
([1243](https://github.com/plotly/plotly.py/pull/1243))
- Added `include_mathjax` argument to `plotly.offline.plot` to support
the creation of HTML files with LaTeX typesetting
([1243](https://github.com/plotly/plotly.py/pull/1243))
- Added new `plotly.offline.get_plotlyjs` function that returns the
contents of the bundled plotly.js library as a string
([637](https://github.com/plotly/plotly.py/issues/637),
[1246](https://github.com/plotly/plotly.py/pull/1246))
- Added new `plotly.offline.get_plotlyjs_version` function that returns
the version of the bundled plotly.js library
([1246](https://github.com/plotly/plotly.py/pull/1246))
- HTML div strings returned by `plotly.offline.plot` now contain logic
to automatically resize the figure responsively. This logic was previously
only added for html files.
([1043](https://github.com/plotly/plotly.py/issues/1043),
[1234](https://github.com/plotly/plotly.py/pull/1234))
- Figures displayed using `plotly.offline.iplot` in the classic Jupyter
Notebook will now resize responsively
([1234](https://github.com/plotly/plotly.py/pull/1234))
- Added `'cdn'`, `'directory'`, and path string `include_plotlyjs` options
in `plotly.offline.plot`
([1234](https://github.com/plotly/plotly.py/pull/1234))
- When `'cdn'`, the resulting html file/div includes a script tag reference
to the plotlyjs cdn.
- When `'directory'`, the resulting html file/div includes a script tag
reference to a plotly.min.js bundle in the same directory as the html file.
If `output_type` is `'file'` then this plotly.min.js bundle is created in
the output directory if it doesn't already exist.
- When a string ending with `'.js'`, the resulting html file/div includes
a script tag that references this exact path. This can be used to point
to a plotly.js bundle from an alternative CDN.
- Added a new `color_threshold` argument to the `create_dendrogram` figure
factory to control the dendrogram clustering cutoff
([995](https://github.com/plotly/plotly.py/issues/995),
[1075](https://github.com/plotly/plotly.py/pull/1075),
[1214](https://github.com/plotly/plotly.py/pull/1214))
- Added support for `autorange='reversed'` in 3D axes
([803](https://github.com/plotly/plotly.py/issues/803),
[plotly/plotly.js3141](https://github.com/plotly/plotly.js/pull/3141))
- Added new gl3d tick and title auto-rotation algorithm that limits text
overlaps
([plotly/plotly.js3084](https://github.com/plotly/plotly.js/pull/3084),
[plotly/plotly.js3131](https://github.com/plotly/plotly.js/pull/3131))
- Added `modebar` layout style attributes:
`orientation`, `bgcolor`, `color` and `activecolor`
([plotly/plotly.js3068](https://github.com/plotly/plotly.js/pull/3068),
[plotly/plotly.js3091](https://github.com/plotly/plotly.js/pull/3091))
- Added `title`, `titleposition` and `titlefont` attributes to pie traces
([plotly/plotly.js2987](https://github.com/plotly/plotly.js/pull/2987))
- Added `hoverlabel.split` attribute to `ohlc` and `candlestick` traces to
split hover labels into multiple pieces
([plotly/plotly.js2959](https://github.com/plotly/plotly.js/pull/2959))
- Added support for `line.shape` values `'hv'`, `'vh'`, `'hvh'`
and `'vhv'` in `scattergl` traces
([plotly/plotly.js3087](https://github.com/plotly/plotly.js/pull/3087))
- Added trace, node and link `hoverinfo` for `sankey` traces
([3096](https://github.com/plotly/plotly.js/pull/3096),
[3150](https://github.com/plotly/plotly.js/pull/3150))
- Added per-sector `textfont` settings in pie traces
([3130](https://github.com/plotly/plotly.js/pull/3130))


Changed
- Use new Plotly logo in "Produced with Plotly" modebar button
([plotly/plotly.js3068](https://github.com/plotly/plotly.js/pull/3068))


Fixed
- Plotly's use of MathJax for LaTeX typesetting no longer interferes with
the Jupyter Notebook's use of MathJax
([445](https://github.com/plotly/plotly.py/issues/445),
[360](https://github.com/plotly/plotly.py/issues/360))
- Fixed several issues with the use of `reversescale=True` in the
`create_annotated_heatmap` figure factory
([1251](https://github.com/plotly/plotly.py/pull/1251))
- Fixed case where `plotly.offline.iplot` would fail to render in the classic
Jupyter Notebook if the notebook contained a Markdown headline with the text
"Plotly"
([816](https://github.com/plotly/plotly.py/issues/816))
- `None` values in a `scatter.hovertext` list are now omitted from the
hover label rather than being displayed as the string `"None"`
([1244](https://github.com/plotly/plotly.py/issues/1244))
- Subplot titles created by `plotly.tools.make_subplots` are now positioned
properly when custom `row_width`/`column_width` arguments are specified
([1229](https://github.com/plotly/plotly.py/issues/1229))
- The `bar.width` property may now be specified as a numpy array or a pandas
series
([1231](https://github.com/plotly/plotly.py/issues/1231),
[plotly/plotly.js3169](https://github.com/plotly/plotly.js/pull/3169))
- Error bars are now scaled correctly for logarithmic `scatter3d` traces
([1139](https://github.com/plotly/plotly.py/issues/1139))
- Use `uuid.uuid4` rather than `uuid.uuid1` to work around an upstream
Python bug
([1235](https://github.com/plotly/plotly.py/issues/1235),
[1236](https://github.com/plotly/plotly.py/pull/1236))
- The `layout.grid.subplots` property may now be specified as a 2D list of
subplot identifiers
([1220](https://github.com/plotly/plotly.py/issues/1220),
[1240](https://github.com/plotly/plotly.py/pull/1240))
- Fixed `scatter3d` text alignment
([1055](https://github.com/plotly/plotly.py/issues/1055),
[plotly/plotly.js3180](https://github.com/plotly/plotly.js/pull/3180))


JupyterLab Versions
For use with JupyterLab, the following versions of the following packages
must be installed:

- Python Packages
- plotly==3.4.0
- ipywidgets>=7.2
- notebook>=5.3
- jupyterlab==0.35

- JupyterLab Extensions
- plotlywidget0.5.0
- jupyter-widgets/jupyterlab-manager0.38
- jupyterlab/plotly-extension0.18

Updated
- Updated Plotly.js to version 1.41.3. Select highlights included below, see
[the plotly.js CHANGELOG](https://github.com/plotly/plotly.js/blob/master/CHANGELOG.md#1413----2018-09-25)
for more information.
- Do not create or check permissions on the `~/.plotly` configuration
directory until a configuration write operation is performed
([1195](https://github.com/plotly/plotly.py/pull/1195)). This change
avoids some concurrency problems associated with running many instances of
plotly.py simultaneously
([1068](https://github.com/plotly/plotly.py/issues/1068)).

Added
- Enable selection by clicking on points via new layout attribute `clickmode` and flag `'select'`
([2944](https://github.com/plotly/plotly.js/pull/2944))
- Added stacked area charts via new attributes `stackgroup` and `stackgaps` in scatter traces
([2960](https://github.com/plotly/plotly.js/pull/2960))
- Added `barpolar` trace type - which replace and augment area traces
([2954](https://github.com/plotly/plotly.js/pull/2954))
- Added `polar.hole` layout parameter to punch hole at the middle of polar
subplot offsetting the start of the radial range
([2977](https://github.com/plotly/plotly.js/pull/2977), [#2996](https://github.com/plotly/plotly.js/pull/2996))
- Figures may now be easily converted to and from JSON using the new
`to_json`, `from_json`, `read_json`, and `write_json` functions in the
`plotly.io` package
([1188](https://github.com/plotly/plotly.py/pull/1188))
- Figures and graph objects now support `deepcopy` and `pickle` operations
([1191](https://github.com/plotly/plotly.py/pull/1191))
- The location of the `"~/.plotly"` settings directory may now be customized
using the `PLOTLY_DIR` environment variable
([1195](https://github.com/plotly/plotly.py/pull/1195))
- Added optional `scaleratio` argument to the `create_quiver` figure factory.
When specified, the axes are restricted to this ratio and the quiver arrows
are computed to have consistent lengths across angles.
([1197](https://github.com/plotly/plotly.py/pull/1197))

Fixed
- Replace use of `pkg_resources.resource_string` with `pkgutil.get_data` to
improve compatibility with `cx_Freeze`
([1201](https://github.com/plotly/plotly.py/pull/1201))
- An exception is no longer raised when an optional dependency raises an
exception on import. The exception is logged and plotly.py continues as if
the dependency were not installed
([1192](https://github.com/plotly/plotly.py/pull/1192))
- Fixed invalid dendrogram axis labels when the points being clustered contain
duplicate values
([1186](https://github.com/plotly/plotly.py/pull/1186))
- Added missing LICENSE.txt file to PyPI source distribution
([765](https://github.com/plotly/plotly.py/issues/765))

JupyterLab Versions
For use with JupyterLab, the following versions of the following packages
must be installed:

- Python Packages
- plotly==3.3.0
- ipywidgets>=7.2
- notebook>=5.3
- jupyterlab==0.34

- JupyterLab Extensions
- plotlywidget0.4.0
- jupyter-widgets/jupyterlab-manager0.37
- jupyterlab/plotly-extension0.17

This is a patch release that fixes a few bugs and reintroduces a few
version 2 features that were not supported in version 3.

The bundled version of plotly.js remains at 1.40.1

JupyterLab Versions
For use with JupyterLab, the following versions of the following packages
must be installed:

- Python Packages
- plotly==3.2.1
- ipywidgets>=7.2
- notebook>=5.3
- jupyterlab==0.34

- JupyterLab Extensions
- plotlywidget0.3.0
- jupyter-widgets/jupyterlab-manager0.37
- jupyterlab/plotly-extension0.17

Added
- An optional `skip_invalid` argument has been added to the `Figure` and
`FigureWidget` constructors. By default, `skip_invalid` is `False` and invalid
figure properties will result in an exception (this is identical to the
previous behavior). When `skip_invalid` is set to `True`, invalid properties
will instead be silently ignored. This argument replaces the `_raise`
argument that was available in version 2, and makes it possible to import
figure definitions from different plotly versions, where incompatible
properties are ignored rather than causing an exception.
- A `to_ordered_dict` method has been added to the `Figure` and `FigureWidget`
classes. This method returns a representation of the figure as a nested
structure of `OrderedDict` and `list` instances where the keys in each
`OrderedDict` are sorted alphabetically. This method replaces the
`get_ordered` method that was available in version 2, and makes it possible
to traverse the nested structure of a figure in a deterministic order.

Fixed
- Pandas `Series` and `Index` objects storing `datetime` values were
incorrectly cast to numeric arrays
([plotly/plotly.py1160](https://github.com/plotly/plotly.py/issues/1160),
[plotly/plotly.py1163](https://github.com/plotly/plotly.py/pull/1163))
- Numpy arrays with `uint64` datatype caused a `FigureWidget` error,
and no figure was displayed
([plotly/plotly.py1155](https://github.com/plotly/plotly.py/issues/1155),
[plotly/plotly.py1163](https://github.com/plotly/plotly.py/pull/1163))