Unexpected initial layout with DataTable and layout()

[ababino]

Dear all,

thanks for this marvelous library! I have written a small dashboard to follow some processes. The dashboard consists of three tables that I render with the layout() function. The problem is that the initial layout has the wrong width making the tables overlap each other.

After moving the browser's window (with the mouse I mean) the elements re-accommodate and they layout as I expected.

In order to reproduce the behavior I saved the code below and I run bokeh serve test.py

test.py file:

from bokeh.layouts import widgetbox, layout
from bokeh.models import ColumnDataSource
from bokeh.models.widgets import DataTable, TableColumn, NumberFormatter
from bokeh.plotting import curdoc
import pandas as pd

df = pd.DataFrame({'Host': ['foo', 'foo', 'bar', 'bar', 'foo', 'foo', 'bar', 'bar'],
                   'JobRuntime': [215, 351, 756, 654, 684, 654, 654, 658],
                   'Exitval': [0, 0, 0, 255, 0, 255, 0, 0]})
source = ColumnDataSource(df)
columns = [
        TableColumn(field="Host", title="Host"),
        TableColumn(field="JobRuntime", title="JobRuntime",
                    formatter=NumberFormatter(format='00:00:00')),
        TableColumn(field="Exitval", title="Exitval"),
    ]
data_table = DataTable(source=source, columns=columns, fit_columns=True,
                       width=400, height=400)

counts = df.pivot_table(columns='Host', index='Exitval', values='JobRuntime',
                        aggfunc=len, margins=True)
counts = counts.reset_index()
counts.columns = counts.columns.map(str)
columns2 = [TableColumn(field=val, title=val) for val in counts.columns]
counts = ColumnDataSource(counts)
data_table2 = DataTable(source=counts, columns=columns2, fit_columns=True,
                        width=800, height=200)

means = df.pivot_table(columns='Host', index='Exitval', values='JobRuntime',
                       margins=True)
means = means.reset_index()
means.columns = means.columns.map(str)
columns3 = [TableColumn(field='Exitval', title='ExitVal')]
columns3 += [TableColumn(field=val, title=val, formatter=NumberFormatter(format='00:00:00')) for val in means.columns[1:]]
means = ColumnDataSource(means)
data_table3 = DataTable(source=means, columns=columns3, fit_columns=True,
                        width=800, height=200)

curdoc().add_root(layout([[widgetbox(data_table), widgetbox(data_table2, data_table3)]], width=3000, sizing_mode='stretch_both'))

The output of bokeh info is
Python version : 2.7.11 |Anaconda 4.0.0 (64-bit)| (default, Dec 6 2015, 18:08:32)
IPython version : 4.1.2
Bokeh version : 0.12.2
BokehJS static path : /home/andres/anaconda2/lib/python2.7/site-packages/bokeh/server/static

Thanks again

[birdsarah]

Thanks for the report and your patience @ababino.

[gryBox]

We also have seen intermittent unexpected behavior when moving the browser window across screens. In our case the bokeh doc loads fine, but when moving the browser across screens, the graph disappears. Reloading the document was our fix :)

Using Windows 7 (64-bit)
python 3.5.x
bokeh 12.1 +

[bryevdv]

With 0.12.6dev7 I now get this from the start:

Perhaps this can be closed?

[bryevdv]

@ababino @gryBox if you have the chance to install and test with 0.12.6dev7 to confirm whether this is resolved it would be appreciated.

cc @mattpap