Extra whitespace with SVG export

[msalvaris]

This was initially reported in the Google groups channel and was encouraged to open an issue here

ALL software version info (bokeh, python, notebook, OS, browser, any other relevant packages)

Ubuntu 16.04
Python 3.6 (Conda env)
Jupyter notebook 5.5
Bokeh 0.13.0
Firefox (although shouldn't matter)

Description of expected behavior and the observed behavior

When exporting a multilevel categorical barchart to SVG it seems to introduce extra whitespace between the levels. This extra whitespace is also apparent in the legend. The effect is worse for longer labels as though the padding is linked to the number of characters in the label.

Complete, minimal, self-contained example code that reproduces the issue

fruits = ['Apples', 'Pears', 'Nectarines', 'Plums', 'Grapes', 'Strawberries']
years = ['2015-01-01', '2016-01-01', '2017-01-01']

data = {'fruits' : fruits,
        '2015-01-01'   : [2, 1, 4, 3, 2, 4],
        '2016-01-01'   : [5, 3, 3, 2, 4, 6],
        '2017-01-01'   : [3, 2, 4, 4, 5, 3]}

# this creates [ ("Apples", "2015"), ("Apples", "2016"), ("Apples", "2017"), ("Pears", "2015), ... ]
x = [ (fruit, year) for fruit in fruits for year in years ]
counts = sum(zip(data['2015-01-01'], data['2016-01-01'], data['2017-01-01']), ()) # like an hstack

year = [i[1] for i in x]

# Normal output
source = ColumnDataSource(data=dict(x=x, counts=counts, year=year))
p = figure(x_range=FactorRange(*x), plot_height=400, plot_width=800, title="Fruit Counts by Year",
           toolbar_location=None, tools="")

p.vbar(x='x', top='counts', width=0.9, source=source, legend='year')

p.y_range.start = 0
p.x_range.range_padding = 0.1
p.xaxis.major_label_orientation = 1.2
p.xgrid.grid_line_color = None

show(p)

#SVG output
source = ColumnDataSource(data=dict(x=x, counts=counts,year=year))
p = figure(x_range=FactorRange(*x), plot_height=400, plot_width=800, title="Fruit Counts by Year",output_backend="svg",
           toolbar_location=None, tools="")

p.vbar(x='x', top='counts', width=0.9, source=source, legend='year')

p.y_range.start = 0
p.x_range.range_padding = 0.1
p.xaxis.major_label_orientation = 1.2
p.xgrid.grid_line_color = None

export_svgs(p, filename="plot.svg")

A notebook and the html output of the notebook can be found here:
https://gist.github.com/msalvaris/3bebd4027a2fe9ba7b428f18ac310ced

[Haptein]

Have you tried the following?
p.background_fill_color = None
p.border_fill_color = None

It is briefly mentioned in the docs that you need to do this in order to get transparent backgrounds. Although it'd be nice if this was the default with svg exports.

[msalvaris]

Hm, no but I don't think my issue is with transparency it is that there is extra whitespace between the levels in the plots when exporting to SVG. This is also apparent in the legend.

[bryevdv]

@canavandl do you have any ideas off the top of your head?

[phausamann]

I would also like to highlight the fact that the axis line width and the ticks are thicker and the last tick seems to be gray instead of black.

This is also independent of the export, any plot with a FactorRange and output_backend='svg' seems to have this problem.