Release notes

Version 1.10.4

This is a minor bug fix release including a number of crucial fixes for issues reported by our users.


  • Allow setting alpha on Image/RGB/HSV and Raster types in bokeh ( #2680 )


  • Fixed bug running display multiple times in one cell ( #2677 )
  • Avoid sending hover data unless explicitly requested ( #2681 )
  • Fixed bug slicing xarray with tuples ( #2674 )

Version 1.10.3

This is a minor bug fix release including a number of crucial fixes for issues reported by our users.


  • The dimensions of elements may now be changed allowing updates to axis labels and table column headers ( #2666 )


  • Fix for labelled plot option ( #2643 )
  • Optimized initialization of dynamic plots specifying a large parameter space ( #2646 )
  • Fixed unicode and reversed axis slicing issues in XArrayInterface ( #2658 , #2653 )
  • Fixed widget sorting issues when applying dynamic groupby ( #2641 )


  • The PlotReset reset parameter was renamed to resetting to avoid clash with a method ( #2665 )
  • PolyDraw tool data parameter now always indexed with ‘xs’ and ‘ys’ keys for consistency ( #2650 )

Version 1.10.2

This is a minor bug fix release with a number of small fixes for features and regressions introduced in 1.10:


  • Exposed Image hover functionality for upcoming bokeh 0.12.16 release ( #2625 )


  • Minor fixes for newly introduced elements and plots including Chord ( #2581 ) and RadialHeatMap ( #2610
  • Fixes for .options method including resolving style and plot option clashes ( #2411 ) and calling it without arguments ( #2630 )
  • Fixes for IPython display function ( #2587 ) and display_formats ( #2592 )


  • BoxWhisker and Bars width bokeh style options and Arrow matplotlib fontsize option are deprecated ( #2411 )

Version 1.10.1

This is a minor bug fix release with a number of fixes for regressions and minor bugs introduced in the 1.10.0 release:


  • Fixed static HTML export of notebooks ( #2574 )
  • Ensured Chord element allows recurrent edges ( #2583 )
  • Restored behavior for inferring key dimensions order from XArray Dataset ( #2579 )
  • Fixed Selection1D stream on bokeh server after changes in bokeh 0.12.15 ( #2586 )

Version 1.10.0

This is a major release with a large number of new features and bug fixes, as well as a small number of API changes. Many thanks to the numerous users who filed bug reports, tested development versions, and contributed a number of new features and bug fixes, including special thanks to @mansenfranzen, @ea42gh, @drs251 and @jakirkham.

JupyterLab support:

  • Full compatibility with JupyterLab when installing the jupyterlab_holoviews extension ( #687 )

New components:


  • Added radial HeatMap option to allow plotting heatmaps with a cyclic x-axis ( #2139 )
  • All elements now support declaring bin edges as well as centers allowing Histogram and QuadMesh to become first class Dataset types ( #547 )
  • When using widgets, their initial or default value can now be set via the Dimension.default parameter ( #704 )
  • n-dimensional Dask arrays are now supported directly via the gridded dictionary data interface ( #2305 )
  • Added new Styling Plots and Colormaps user guides, including new functionality for working with colormaps.


  • Improvements to exceptions ( #1127 )
  • Toolbar position and merging (via a new merge_toolbar option) can now be controlled for Layout and Grid plots ( #1977 )
  • Bokeh themes can now be applied at the renderer level ( #1861 )
  • Dataframe and Series index can now be referenced by name when constructing an element ( #2000 )
  • Option-setting methods such as .opts , .options and hv.opts now allow specifying the backend instead of defaulting to the current backend ( #1801 )
  • Handled API changes in streamz 0.3.0 in Buffer stream ( #2409 )
  • Supported GIF output on windows using new Matplotlib pillow animation support ( #385 )
  • Provided simplified interface to rasterize most element types using datashader ( #2465 )
  • Bivariate element now support levels as a plot option ( #2099 )
  • NdLayout and GridSpace now consistently support * overlay operation ( #2075 )
  • The Bokeh backend no longer has a hard dependency on Matplotlib ( #829 )
  • DynamicMap may now return ( Nd ) Overlay with varying number of elements ( #1388 )
  • In the notebook, deleting or re-executing a cell will now delete the plot and clean up any attached streams ( #2141 )
  • Added color_levels plot option to set discrete number of levels during colormapping ( #2483 )
  • Expanded the Large Data user guide to show examples of all Element and Container types supported for datashading and give performance guidelines.


  • Layout and Overlay objects no longer create lower-case nodes on attribute access ( #2331 )
  • Dimension.step now correctly respects both integer and float steps ( #1707 )
  • Fixed timezone issues when using linked streams on datetime axes ( #2459 )

Changes affecting backwards compatibility:

  • Image elements now expect and validate regular sampling ( #1869 ); for genuinely irregularly sampled data QuadMesh should be used.
  • Tabular elements will no longer default to use ArrayInterface , instead preferring pandas and dictionary data formats ( #1236 )
  • Cycle / Palette values are no longer zipped together; instead they now cycle independently ( #2333 )
  • The default color Cycle was expanded to provide more unique colors ( #2483 )
  • Categorical colormapping was made consistent across backends, changing the behavior of categorical Matplotlib colormaps ( #2483 )
  • Disabled auto-indexable property of the Dataset baseclass, i.e. if a single column is supplied no integer index column is added automatically ( #2522 )

Version 1.9

Version 1.9.5

This release includes a very small number of minor bugfixes and a new feature to simplify setting options in python:


  • Added .options method for simplified options setting. ( #2306 )


  • Allow plotting bytes datausing the Bokeh backend in python3 ( #2357 )
  • Allow .range to work on data with heterogeneous types in Python 3 ( #2345 )
  • Fixed bug streaming data containing datetimes using bokeh>-0.12.14 ( #2383 )

Version 1.9.4

This release contains a small number of important bug fixes:

  • Compatibility with recent versions of Dask and pandas ( #2329 )
  • Fixed bug referencing columns containing non-alphanumeric characters in Bokeh Tables ( #2336 )
  • Fixed issue in regrid operation ( 2337 )
  • Fixed issue when using datetimes with datashader when processing ranges ( #2344 )

Version 1.9.3

This release contains a number of important bug fixes and minor enhancements.

Particular thanks to @jbampton, @ea42gh, @laleph, and @drs251 for a number of fixes and improvements to the documentation.


  • Optimized rendering of stream based OverlayPlots ( #2253 )
  • Added merge_toolbars and toolbar options to control toolbars on Layout and Grid plots ( #2289 )
  • Optimized rendering of VectorField ( #2314 )
  • Improvements to documentation ( #2198 , #2220 , #2233 , #2235 , #2316 )
  • Improved Bokeh Table formatting ( #2267 )
  • Added support for handling types ( #2267 )
  • Add support for pre- and post-process hooks on operations ( #2246 , #2334 )


  • Fix for Bokeh server widgets ( #2218 )
  • Fix using event based streams on Bokeh server ( #2239 , #2256 )
  • Switched to drawing Distribution , Area and Spread using patch glyphs in Bokeh fixing legends ( #2225 )
  • Fixed categorical coloring of Polygons / Path elements in Matplotlib ( #2259 )
  • Fixed bug computing categorical datashader aggregates ( #2295 )
  • Allow using Empty object in AdjointLayout ( #2275 )

API Changes:

  • Renamed Trisurface to TriSurface for future consistency ( #2219 )

Version 1.9.2

This release is a minor bug fix release patching various issues which were found in the 1.9.1 release.


  • Improved the Graph element, optimizing the constructor and adding support for defining a edge_color_index ( #2145 )
  • Added support for adding jitter to Bokeh Scatter and Points plots ( e56208 )


  • Ensure dimensions, group and label are inherited when casting Image to QuadMesh ( #2144 )
  • Handle compatibility for Bokeh version >- 0.12.11 ( #2159 )
  • Fixed broken Bokeh ArrowPlot ( #2172 )
  • Fixed Pointer based streams on datetime axes ( #2179 )
  • Allow constructing and plotting of empty Distribution and Bivariate elements ( #2190 )
  • Added support for hover info on Bokeh BoxWhisker plots ( #2187 )
  • Fixed bug attaching streams to (Nd)Overlay types ( #2194 )

Version 1.9.1

This release is a minor bug fix release patching various issues which were found in the 1.9.0 release.


  • Exposed min_alpha parameter on datashader shade and datashade operations ( #2109 )


  • Fixed broken Bokeh server linked stream throttling ( #2112 )
  • Fixed bug in Bokeh callbacks preventing linked streams using Bokeh’s on_event callbacks from working ( #2112 )
  • Fixed insufficient validation issue for Image and bugs when applying regrid operation to xarray based Images ( #2117 )
  • Fixed handling of dimensions and empty elements in univariate_kde and bivariate_kde operations ( #2103 )

Version 1.9.0

This release includes a large number of long awaited features, improvements and bug fixes, including streaming and graph support, binary transfer of Bokeh data, fast Image/RGB regridding, first-class statistics elements and a complete overhaul of the geometry elements.

Particular thanks to all users and contributers who have reported issues and submitted pull requests.


  • The kdim and vdim keyword arguments are now positional making the declaration of elements less verbose (e.g. Scatter(data, ‘x’, ‘y’)) ( #1946 )
  • Added Graph, Nodes, and EdgePaths elements adding support for plotting network graphs ( #1829 )
  • Added datashader based regrid operation for fast Image and RGB regridding ( #1773 )
  • Added support for binary transport when plotting with Bokeh, providing huge speedups for dynamic plots ( #1894 , #1896 )
  • Added Pipe and Buffer streams for streaming data support ( #2011 )
  • Add support for datetime axes on Image, RGB and when applying datashading and regridding operations ( #2023 )
  • Added Distribution and Bivariate as first class elements which can be plotted with Matplotlib and Bokeh without depending on seaborn ( #1985 )
  • Completely overhauled support for plotting geometries with Path, Contours and Polygons elements including support for coloring individual segments and paths by value ( #1991 )


  • Add support for adjoining all elements on Matplotlib plots ( #1033 )
  • Improved exception handling for data interfaces ( #2041 )
  • Add groupby argument to histogram operation ( #1725 )
  • Add support for reverse sort on Dataset elements ( #1843 )
  • Added support for invert_x/yaxis on all elements ( #1872 , #1919 )


  • Fixed a bug in Matplotlib causing the first frame in gif and mp4 getting stuck ( #1922 )
  • Fixed various issues with support for new nested categorical axes in Bokeh ( #1933 )
  • A large range of other bug fixes too long to list here.

Changes affecting backwards compatibility:

  • The contours operation no longer overlays the contours on top of the supplied Image by default and returns a single Contours/Polygons rather than an NdOverlay of them ( #1991 )
  • The values of the Distribution element should now be defined as a key dimension ( #1985 )
  • The seaborn interface was removed in its entirety being replaced by first class support for statistics elements such as Distribution and Bivariate ( #1985 )
  • Since kdims and vdims can now be passed as positional arguments the bounds argument on Image is no longer positional ( #1946 ).
  • The datashade and shade cmap was reverted back to blue due to issues with the fire cmap against a white background. ( #2078 )
  • Dropped all support for Bokeh versions older than 0.12.10
  • histogram operation now returns Histogram elements with less generic value dimension and customizable label ( #1836 )

Version 1.8.4

This bugfix release includes a number of critical fixes for compatiblity with Bokeh 0.12.9 along with various other bug fixes. Many thanks to our users for various detailed bug reports, feedback and contributions.


  • Fixes to register BoundsXY stream. ( #1826 )
  • Fix for Bounds streams on Bokeh server. ( #1883 )
  • Compatibility with Matplotlib 2.1 ( #1842 )
  • Fixed bug in scrubber widget and support for scrubbing discrete DynamicMaps ( #1832 )
  • Various fixes for compatibility with Bokeh 0.12.9 ( #1849 , #1866 )
  • Fixes for setting QuadMesh ranges. ( #1876 )
  • Fixes for inverting Image/RGB/Raster axes in Bokeh. ( #1872 )

Version 1.8.3

This bugfix release fixes a number of minor issues identified since the last release:


  • Add support for setting the Bokeh sizing_mode as a plot option ( #1813 )


  • Handle StopIteration on DynamicMap correctly. ( #1792 )
  • Fix bug with linked streams on empty source element ( #1725 )
  • Compatibility with latest datashader 0.6.0 release ( #1773 )
  • Fixed missing HTML closing tag in extension ( #1797 , #1809 )
  • Various fixes and improvements for documentation ( #1664 , #1796 )

Version 1.8.2

This bugfix release addresses a number of minor issues identified since the 1.8.1 release:


  • Added support for groupby to histogram operation. ( #1725 )


Version 1.8.1

This bugfix release addresses a number of minor issues identified since the 1.8 release:


  • All enabled plotting extension logos now shown ( #1694 )


  • Updated search ordering when looking for holoviews.rc ( #1700 )
  • Fixed lower bound inclusivity bug when no upper bound supplied ( #1686 )
  • Raise SkipRendering error when plotting nested layouts ( #1687 )
  • Added safety margin for grid axis constraint issue ( #1695 )
  • Fixed bug when using +framewise ( #1685 )
  • Fixed handling of Spacer models in sparse grid ( #1682 )
  • Renamed Bounds to BoundsXY for consistency ( #1672 )
  • Fixed Bokeh log axes with axis lower bound <-0 ( #1691 )
  • Set default datashader cmap to fire ( #1697 )
  • Set SpikesPlot color index to None by default ( #1671 )
  • Documentation fixes ( #1662 , #1665 , #1690 , #1692 , #1658 )

Version 1.8.0

This release includes a complete and long awaited overhaul of the HoloViews documentation and website, with a new gallery, getting-started section, and logo. In the process, we have also improved and made small fixes to all of the major new functionality that appeared in 1.7.0 but was not properly documented until now. We want to thank all our old and new contributors for providing feedback, bug reports, and pull requests.

Major features:

  • Completely overhauled the documentation and website ( #1384 , #1473 , #1476 , #1473 , #1537 , #1585 , #1628 , #1636 )
  • Replaced dependency on bkcharts with new Bokeh bar plot ( #1416 ) and Bokeh BoxWhisker plot ( #1604 )
  • Added support for drawing the Arrow annotation in Bokeh ( #1608 )
  • Added periodic method DynamicMap to schedule recurring events ( #1429 )
  • Cleaned up the API for deploying to Bokeh server ( #1444 , #1469 , #1486 )
  • Validation of invalid backend specific options ( #1465 )
  • Added utilities and entry points to convert notebooks to scripts including magics ( #1491 )
  • Added support for rendering to png in Bokeh backend ( #1493 )
  • Made Matplotlib and Bokeh styling more consistent and dropped custom Matplotlib rc file ( #1518 )
  • Added iloc and ndloc method to allow integer based indexing on tabular and gridded datasets ( #1435 )
  • Added option to restore case sensitive completion order by setting hv.extension.case_sensitive_completion-True in python or via holoviews.rc file ( #1613 )

Other new features and improvements:

Changes affecting backwards compatibility:

  • Renamed ElementOperation to Operation ( #1421 )
  • Removed stack_area operation in favor of Area.stack classmethod ( #1515 )
  • Removed all mpld3 support ( #1516 )
  • Added opts method on all types, replacing the now-deprecated __call__ syntax to set options ( #1589 )
  • Styling changes for both Matplotlib and Bokeh, which can be reverted for a notebook with the config option of hv.extension . For instance, hv.extension('bokeh', config-dict(style_17-True)) ( #1518 )

Version 1.7.0

This version is a major new release incorporating seven months of work involving several hundred PRs and over 1700 commits. Highlights include extensive new support for easily building highly interactive Bokeh plots, support for using datashader -based plots for working with large datasets, support for rendering images interactively but outside of the notebook, better error handling, and support for Matplotlib 2.0 and Bokeh 0.12.5. The PRs linked below serve as initial documentation for these features, and full documentation will be added in the run-up to HoloViews 2.0.

Major features and improvements:

Other new features and improvements:

Changes affecting backwards compatibility:

  • Automatic coloring and sizing on Points now disabled (PR #748 ).
  • Deprecated max_branches output magic option (PR #1293 ).
  • Deprecated GridImage (PR #1292 , #1223 ).
  • Deprecated NdElement (PR #1191 ).
  • Deprecated DFrame conversion methods (PR #1065 ).
  • Banner text removed from notebook_extension() (PR #1231 , #1291 ).
  • Bokeh’s Matplotlib compatibility module removed (PR #1239 ).
  • ls as Matplotlib linestyle alias dropped (PR #1203 ).
  • mdims argument of conversion interface renamed to groupby (PR #1066 ).
  • Replaced global alias state with Dimension.label ( #1083 ).
  • DynamicMap only update ranges when set to framewise
  • Deprecated DynamicMap sampled, bounded, open and generator modes ( #969 , #1305 )
  • Layout.display method is now deprecated ( #1026 )
  • Layout fix for Matplotlib figures with non-square aspects introduced in 1.6.2 (PR #826 ), now enabled by default.

Version 1.6.2

Bug fix release with various fixes for gridded data backends and optimizations for Bokeh.

  • Optimized Bokeh event messaging, reducing the average json payload by 30-50% (PR #807 ).
  • Fixes for correctly handling NdOverlay types returned by DynamicMaps (PR #814 ).
  • Added support for datetime64 handling in Matplotlib and support for datetime formatters on Dimension.type_formatters (PR #816 ).
  • Fixed handling of constant dimensions when slicing xarray datasets (PR #817 ).
  • Fixed support for passing custom dimensions to iris Datasets (PR #818 ).
  • Fixed support for add_dimension on xarray interface (PR #820 ).
  • Improved extents computation on Matplotlib SpreadPlot (PR #821 ).
  • Bokeh backend avoids sending data for static frames and empty events (PR #822 ).
  • Added major layout fix for figures with non-square aspects, reducing the amount of unnecessary whitespace (PR #826 ). Disabled by default until 1.7 release but can be enabled with:
from holoviews.plotting.mpl import LayoutPlot
LayoutPlot.v17_layout_format - True
LayoutPlot.vspace - 0.3

Version 1.6.1

Bug fix release following the 1.6 major release with major bug fixes for the grid data interfaces and improvements to the options system.

  • Ensured that style options incompatible with active backend are ignored (PR #802 ).
  • Added support for placing legends outside the plot area in Bokeh (PR #801 ).
  • Fix to ensure Bokeh backend does not depend on pandas (PR #792 ).
  • Fixed option system to ensure correct inheritance when redefining options (PR #796 ).
  • Major refactor and fixes for the grid based data backends (iris, xarray and arrays with coordinates) ensuring the data is oriented and transposed correctly (PR #794 ).

Version 1.6

A major release with an optional new data interface based on xarray, support for batching Bokeh plots for huge increases in performance, support for Bokeh 0.12 and various other fixes and improvements.

Features and improvements:

  • Made VectorFieldPlot more general with support for independent coloring and scaling (PR #701 ).
  • Iris interface now allows tuple and dict formats in the constructor (PR #709 .
  • Added support for dynamic groupby on all data interfaces (PR #711 ).
  • Added an xarray data interface (PR #713 ).
  • Added the redim method to all Dimensioned objects making it easy to quickly change dimension names and attributes on nested objects #715 ).
  • Added support for batching plots (PR #715 ).
  • Support for Bokeh 0.12 release (PR #725 ).
  • Added support for logz option on Bokeh Raster plots (PR #729 ).
  • Bokeh plots now support custom tick formatters specified via Dimension value_format (PR #728 ).

Version 1.5

A major release with a large number of new features including new data interfaces for grid based data, major improvements for DynamicMaps and a large number of bug fixes.

Features and improvements:

  • Added a grid based data interface to explore n-dimensional gridded data easily (PR #562 ).
  • Added data interface based on iris Cubes (PR #624 ).
  • Added support for dynamic operations and overlaying of DynamicMaps (PR #588 ).
  • Added support for applying groupby operations to DynamicMaps (PR #667 ).
  • Added dimension value formatting in widgets (PR #562 ).
  • Added support for indexing and slicing with a function (PR #619 ).
  • Improved throttling behavior on widgets (PR #596 ).
  • Major refactor of Matplotlib plotting classes to simplify implementing new Element plots (PR #438 ).
  • Added Renderer.last_plot attribute to allow easily debugging or modifying the last displayed plot (PR #538 ).
  • Added Bokeh QuadMeshPlot (PR #661 ).

Bug fixes:

  • Fixed overlaying of 3D Element types (PR #504 ).
  • Fix for Bokeh hovertools with dimensions with special characters (PR #524 ).
  • Fixed bugs in seaborn Distribution Element (PR #630 ).
  • Fix for inverted Raster.reduce method (PR #672 ).
  • Fixed Store.add_style_opts method (PR #587 ).
  • Fixed bug preventing simultaneous logx and logy plot options (PR #554 ).

Backwards compatibility:

  • Renamed Columns type to Dataset (PR #620 ).

Version 1.4.3

A minor bugfix release to patch a number of small but important issues.

Fixes and improvements:

  • Added a DynamicMap Tutorial to explain how to explore very large or continuous parameter spaces in HoloViews ( PR #470 ).
  • Various fixes and improvements for DynamicMaps including slicing ( PR #488 ) and validation ( PR #483 ) and serialization ( PR #483 )
  • Widgets containing Matplotlib plots now display the first frame from cache providing at least the initial frame when exporting DynamicMaps ( PR #486 )
  • Fixed plotting Bokeh plots using widgets in live mode, after changes introduced in latest Bokeh version (commit 1b87c91e9 ).
  • Fixed issue in coloring Point/Scatter objects by values ( Issue #467 ).

Backwards compatibility:

  • The behavior of the scaling_factor on Point and Scatter plots has changed now simply multiplying area or width (as defined by the scaling_method ). To disable scaling points by a dimension set size_index-None .
  • Removed hooks to display 3D Elements using the BokehMPLRawWrapper in Bokeh ( PR #477 )
  • Renamed the DynamicMap mode closed to bounded ( PR #477 )

Version 1.4.2

Over the past month since the 1.4.1 release, we have improved our infrastructure for building documentation, updated the main website and made several additional usability improvements.

Documentation changes:

  • Major overhaul of website and notebook building making it much easier to test user contributions ( Issue #180 , PR #429 )
  • Major rewrite of the documentation ( PR #401 , PR #411 )
  • Added Columnar Data Tutorial and removed most of Pandas Conversions as it is now supported by the core.

Fixes and improvements:

  • Major improvement for grid based layouts with varying aspects ( PR #457 )
  • Fix for interleaving %matplotline inline and holoviews plots ( Issue #179 )
  • Matplotlib legend z-orders and updating fixed ( Issue #304 , Issue #305 )
  • color_index and size_index plot options support specifying dimension by name ( Issue #391 )
  • Added Area Element type for drawing area under or between Curves. ( PR #427 )
  • Fixed issues where slicing would remove styles applied to an Element. ( Issue #423 , PR #439 )
  • Updated the title_format plot option to support a {dimensions} formatter ( PR #436 )
  • Improvements to Renderer API to allow JS and CSS requirements for exporting standalone widgets ( PR #426 )
  • Compatibility with the latest Bokeh 0.11 release ( PR #393 )

Version 1.4.1

Over the past two weeks since the 1.4 release, we have implemented several important bug fixes and have made several usability improvements.

New features:

  • Improved help system. It is now possible to recursively list all the applicable documentation for a composite object. In addition, the documentation may now be filtered using a regular expression pattern. ( PR #370 )
  • HoloViews now supports multiple active display hooks making it easier to use nbconvert. For instance, PNG data will be embedded in the notebook if the argument display_formats-[‘html’,’png’] is supplied to the notebook_extension. ( PR #355 )
  • Improvements to the display of DynamicMaps as well as many new improvements to the Bokeh backend including better VLines/HLines and support for the Bars element. ( PR #367 , PR #362 , PR #339 ).
  • New Spikes and BoxWhisker elements suitable for representing distributions as a sequence of lines or as a box-and-whisker plot. ( PR #346 , PR #339 )
  • Improvements to the notebook_extension. For instance, executing hv.notebook_extension(‘bokeh’) will now load BokehJS and automatically activate the Bokeh backend (if available).
  • Significant performance improvements when using the groupby operation on HoloMaps and when working with highly nested datastructures. ( PR #349 , PR #359 )

Notable bug fixes:

  • DynamicMaps are now properly integrated into the style system and can be customized in the same way as HoloMaps. ( PR #368 )
  • Widgets now work correctly when unicode is used in the dimension labels and values ( PR #376 ).

Version 1.4.0

Over the past few months we have added several major new features and with the help of our users have been able to address a number of bugs and inconsistencies. We have closed 57 issues and added over 1100 new commits.

Major new features:

  • Data API: The new data API brings an extensible system of to add new data interfaces to column based Element types. These interfaces allow applying powerful operations on the data independently of the data format. The currently supported datatypes include NumPy, pandas dataframes and a simple dictionary format. ( PR #284 )
  • Backend API: In this release we completely refactored the rendering, plotting and IPython display system to make it easy to add new plotting backends. Data may be styled and pickled for each backend independently and renderers now support exporting all plotting data including widgets as standalone HTML files or with separate JSON data.
  • Bokeh backend: The first new plotting backend added via the new backend API. Bokeh plots allow for much faster plotting and greater interactivity. Supports most Element types and layouts and provides facilities for sharing axes across plots and linked brushing across plots. ( PR #250 )
  • DynamicMap: The new DynamicMap class allows HoloMap data to be generated on-the-fly while running a Jupyter IPython notebook kernel. Allows visualization of unbounded data streams and smooth exploration of large continuous parameter spaces. ( PR #278 )

Other features:

  • Easy definition of custom aliases for group, label and Dimension names, allowing easier use of LaTeX.
  • New Trisurface and QuadMesh elements.
  • Widgets now allow expressing hierarchical relationships between dimensions.
  • Added GridMatrix container for heterogeneous Elements and gridmatrix operation to generate scatter matrix showing relationship between dimensions.
  • Filled contour regions can now be generated using the contours operation.
  • Consistent indexing semantics for all Elements and support for boolean indexing for Columns and NdMapping types.
  • New hv.notebook_extension function offers a more flexible alternative to %load_ext, e.g. for loading other extensions hv.notebook_extension(bokeh-True).

Experimental features:

  • Bokeh callbacks allow adding interactivity by communicating between BokehJS tools and the IPython kernel, e.g. allowing downsampling based on the zoom level.

Notable bug fixes:

  • Major speedup rendering large HoloMaps (~ 2-3 times faster).
  • Colorbars now consistent for all plot configurations.
  • Style pickling now works correctly.

API Changes:

  • Dimension formatter parameter now deprecated in favor of value_format.
  • Types of Chart and Table Element data now dependent on selected interface.
  • DFrame conversion interface deprecated in favor of Columns pandas interface.

Version 1.3.2

Minor bugfix release to address a small number of issues:


  • Added support for colorbars on Surface Element (1cd5281).
  • Added linewidth style option to SurfacePlot (9b6ccc5).

Bug fixes:

  • Fixed inversion inversion of y-range during sampling (6ff81bb).
  • Fixed overlaying of 3D elements (787d511).
  • Ensuring that underscore.js is loaded in widgets (f2f6378).
  • Fixed Python3 issue in Overlay.get (8ceabe3).

Version 1.3.1

Minor bugfix release to address a number of issues that weren’t caught in time for the 1.3.0 release with the addition of a small number of features:


  • Introduced new Spread element to plot errors and confidence intervals (30d3184).
  • ErrorBars and Spread elements now allow most Chart constructor types (f013deb).

Bug fixes:

  • Fixed unicode handling for dimension labels (061e9af).
  • Handling of invalid dimension label characters in widgets (a101b9e).
  • Fixed setting of fps option for MPLRenderer video output (c61b9df).
  • Fix for multiple and animated colorbars (5e1e4b5).
  • Fix to Chart slices starting or ending at zero (edd0039).

Version 1.3.0

Since the last release we closed over 34 issues and have made 380 commits mostly focused on fixing bugs, cleaning up the API and working extensively on the plotting and rendering system to ensure HoloViews is fully backend independent.

We’d again like to thank our growing user base for all their input, which has helped us in making the API more understandable and fixing a number of important bugs.


  • Allowed display of data structures which do not match the recommended nesting hierarchy (67b28f3, fbd89c3).
  • Dimensions now sanitized for .select , .sample and .reduce calls (6685633, 00b5a66).
  • Added holoviews.ipython.display function to render (and display) any HoloViews object, useful for IPython interact widgets (0fa49cd).
  • Table column widths now adapt to cell contents (be90a54).
  • Defaulting to Matplotlib ticking behavior (62e1e58).
  • Allowed specifying fixed figure sizes to Matplotlib via fig_inches tuples using (width, None) and (None, height) formats (632facd).
  • Constructors of Chart , Path and Histogram classes now support additional data formats (2297375).
  • ScrubberWidget now supports all figure formats (c317db4).
  • Allowed customizing legend positions on Bars Elements (5a12882).
  • Support for multiple colorbars on one axis (aac7b92).
  • .reindex on NdElement types now support converting between key and value dimensions allowing more powerful conversions. (03ac3ce)
  • Improved support for casting between Element types (cdaab4e, b2ad91b, ce7fe2d, 865b4d5).
  • The %%opts cell magic may now be used multiple times in the same cell (2a77fd0)
  • Matplotlib rcParams can now be set correctly per figure (751210f).
  • Improved OptionTree repr which now works with eval (2f824c1).
  • Refactor of rendering system and IPython extension to allow easy swapping of plotting backend (#141)
  • Large plotting optimization by computing tight bbox_inches once (e34e339).
  • Widgets now cache frames in the DOM, avoiding flickering in some browsers and make use of jinja2 template inheritance. (fc7dd2b)
  • Calling a HoloViews object without arguments now clears any associated custom styles. (9e8c343)

API Changes

  • Renamed key_dimensions and value_dimensions to kdims and vdims respectively, while providing backward compatibility for passing and accessing the long names (8feb7d2).
  • Combined x/y/zticker plot options into x/y/zticks parameters which now accept an explicit number of ticks, an explicit list of tick positions (and labels), and a Matplotlib tick locator.
  • Changed backend options in %output magic, nbagg and d3 are now modes of the Matplotlib backend and can be selected with backend-'matplotlib:nbagg' and backend-'matplotlib:mpld3' respectively. The ‘d3’ and ‘nbagg’ options remain supported but will be deprecated in future.
  • Customizations should no longer be applied directly to Store.options ; the Store.options(backend-'matplotlib') object should be customized instead. There is no longer a need to call the deprecated Store.register_plots method.

Version 1.2.0

Since the last release we closed over 20 issues and have made 334 commits, adding a ton of functionality and fixing a large range of bugs in the process.

In this release we received some excellent feedback from our users, which has been greatly appreciated and has helped us address a wide range of problems.


  • Added new ErrorBars Element (f2b276b)
  • Added Empty pseudo-Element to define empty placeholders in Layouts (35bac9f1d)
  • Added support for changing font sizes easily (0f54bea)
  • Support for holoviews.rc file (79076c8)
  • Many major speed optimizations for working with and plotting HoloViews data structures (fe87b4c, 7578c51, 5876fe6, 8863333)
  • Support for GridSpace with inner axes (93295c8)
  • New aspect_weight and tight Layout plot options for more customizability of Layout arrangements (4b1f03d, e6a76b7)
  • Added bgcolor plot option to easily set axis background color (92eb95c)
  • Improved widget layout (f51af02)
  • New OutputMagic css option to style html output (9d42dc2)
  • Experimental support for PDF output (1e8a59b)
  • Added support for 3D interactivity with nbagg (781bc25)
  • Added ability to support deprecated plot options in %%opts magic.
  • Added DrawPlot simplifying the implementation of custom plots (38e9d44)

API changes:

  • Path and Histogram support new constructors (7138ef4, 03b5d38)

  • New depth argument on the relabel method (f89b89f)

  • Interface to Pandas improved (1a7cd3d)

  • Removed xlim , ylim and zlim to eliminate redundancy.

  • Renaming of various plot and style options including:

    • figure_* to fig_*
    • vertical_spacing and horizontal_spacing to vspace and hspace respectively

    * Deprecation of confusing origin style option on RasterPlot

  • Overlay.__getitem__ no longer supports integer indexing (use get method instead)

Important bug fixes:

  • Important fixes to inheritance in the options system (d34a931, 71c1f3a7)
  • Fixes to the select method (df839bea5)
  • Fixes to normalization system (c3ef40b)
  • Fixes to Raster and Image extents, __getitem__ and sampling.
  • Fixed bug with disappearing adjoined plots (2360972)
  • Fixed plot ordering of overlaid elements across a HoloMap (c4f1685)

Version 1.1.0


  • Support for nbagg as a backend (09eab4f1)
  • New .hvz file format for saving HoloViews objects (bfd5f7af)
  • New Polygon element type (d1ec8ec8)
  • Greatly improved Unicode support throughout, including support for unicode characters in Python 3 attribute names (609a8454)
  • Regular SelectionWidget now supports live rendering (eb5bf8b6)
  • Supports a list of objects in Layout and Overlay constructors (5ba1866e)
  • Polar projections now supported (3801b76e)

API changes (not backward compatible):

  • xlim , ylim , zlim , xlabel , ylabel and zlabel have been deprecated (081d4123)
  • Plotting options show_xaxis and show_yaxis renamed to xaxis and yaxis , respectively (13393f2a).
  • Deprecated IPySelectionWidget (f59c34c0)

In addition to the above improvements, many miscellaneous bug fixes were made.

Version 1.0.1

Minor release addressing bugs and issues with 1.0.0.


  • New separate Pandas Tutorial (8455abc3)
  • Silenced warnings when loading the IPython extension in IPython 3 (aaa6861b)
  • Added more useful installation options via (72ece4db)
  • Improvements and bug-fixes for the %%opts magic tab-completion (e0ad7108)
  • DFrame now supports standard constructor for pandas dataframes (983825c5)
  • Tables are now correctly formatted using the appropriate Dimension formatter (588bc2a3)
  • Support for unlimited alphabetical subfigure labelling (e039d00b)
  • Miscellaneous bug fixes, including Python 3 compatibility improvements.

Version 1.0.0

First public release available on GitHub and PyPI.