Getting back toolbar on left (if set to top position) in 0.8.4?

[sdbbs]

Describe the issue

In an earlier version of ipympl (unfortunately, haven't recorded which), if I chose the positioning of the toolbar on top, it was automatically placed on the left; I have a partial record of that here - https://stackoverflow.com/questions/69106967/jupyter-matplotlib-widget-place-toolbar-horizontally-over-plot

However, I made an update recently, and it turns out, the hacks that I have documented previously, do not work anymore - also because now the default for fig.canvas.toolbar_visible seems to be fade-in-fade-out (apparently, when I documented above hacks, the default was visible).

So now, if you run the last code snippet in the answer of the linked SO post above (only thing I added was version printout):

%matplotlib widget
import ipympl; print("ipympl version {}".format(ipympl.__version__))

import matplotlib.pyplot as plt
from ipywidgets import widgets, Layout
from IPython.display import display, HTML

out_widget_plot = widgets.Output()
with out_widget_plot:
    with plt.style.context("default"): # ggplot
        fig, (ax1) = plt.subplots(1, 1)

fig.suptitle('test title', fontsize=10) 
tline = ax1.plot([0,1,2], [10,20,30])
fig.tight_layout()
#fig.canvas.toolbar.button_style = 'info' # test: ok, works (but don't need it here)
#fig.canvas.toolbar.orientation = 'horizontal' # no error, but no effect either
fig.canvas.toolbar_position = 'top' # this also switches the orientation to horizontal, too
fig.canvas.header_visible = False # Gets rid of "Figure 1" on top

# the below CSS style override in Jupyter, allows the div that hosts the Matplotlib toolbar, to be placed "on top" of the div that hosts the plot:
# widget-label does the same, but for the bottom "x/y" indicator (which is why its position needs to also be controlled with margin)
sthtml = """
<style>
.jupyter-matplotlib-toolbar {
    overflow: visible;
    position: absolute;
    z-index: 1000;
}
.widget-label {
    color: var(--jp-widgets-label-color);
    font-size: var(--jp-widgets-font-size);
    overflow: hidden;
    text-overflow: ellipsis;
    white-space: nowrap;
    line-height: var(--jp-widgets-inline-height);
    z-index: 1000;
    margin-top: -41px;
    margin-left: 58px;
}
</style>
"""

the_ui = widgets.VBox([ widgets.HTML(sthtml), widgets.HTML("<b>Hello</b>"), out_widget_plot, widgets.HTML("<b>World</b>"),])
display(the_ui)

... you at first do not get any toolbar menu (because due to the fade-in-fade-out, the menu is not shown, if the mouse cursor is not over the plot widget), then when you mouseover, you get the toolbar to the right (and there is no "hamburger" icon at first, that used to collapse/expand the rest of the toolbar; and the plot figure rendered after the two HTML "labels", "Hello" and "World"):

... whereas the same code earlier showed a toolbar on the left (and there is "hamburger" icon at first, that used to collapse/expand the rest of the toolbar, shown; and the plot being rendered between the two HTML "labels") :

Note also that now, the figure plot widget is not shown anymore between the "Hello" and the "World" labels - whereas previously it was (which was the intention).

Is there any way I can get the old behavior back, so I can place the toolbar on the top left, and have the figure shown between other widgets (and possibly get the "hamburger" expand/collapse button back)?

Versions

3.8.10 (default, Sep 28 2021, 16:10:42) 
[GCC 9.3.0]
ipympl version: 0.8.4
Selected Jupyter core packages...
IPython          : 7.30.1
ipykernel        : 6.6.0
ipywidgets       : 7.6.5
jupyter_client   : 7.1.0
jupyter_core     : 4.9.1
jupyter_server   : 1.13.1
jupyterlab       : not installed
nbclient         : 0.5.9
nbconvert        : 6.3.0
nbformat         : 5.1.3
notebook         : 6.4.6
qtconsole        : 5.2.2
traitlets        : 5.1.1
Known nbextensions:
  config dir: /usr/local/etc/jupyter/nbconfig
    notebook section
      bqplot/extension  enabled 
      - Validating: OK
      ipysheet/extension  enabled 
      - Validating: OK
      ipyvolume/extension  enabled 
      - Validating: OK
      jupyter-datawidgets/extension  enabled 
      - Validating: OK
      jupyter-leaflet/extension  enabled 
      - Validating: OK
      jupyter-matplotlib/extension  enabled 
      - Validating: OK
      jupyter-threejs/extension  enabled 
      - Validating: OK
      jupyter-webrtc/extension  enabled 
      - Validating: OK
      jupyter_bokeh/extension  enabled 
      - Validating: OK
      jupyter_resource_usage/main  enabled 
      - Validating: OK
      jupyterlab-plotly/extension  enabled 
      - Validating: OK
      voila/extension  enabled 
      - Validating: OK
      jupyter-js-widgets/extension  enabled 
      - Validating: OK
      nbextensions_configurator/config_menu/main  enabled 
      - Validating: problems found:
        - require?  X nbextensions_configurator/config_menu/main
      contrib_nbextensions_help_item/main  enabled 
      - Validating: OK
    tree section
      nbextensions_configurator/tree_tab/main  enabled 
      - Validating: problems found:
        - require?  X nbextensions_configurator/tree_tab/main
usage: jupyter [-h] [--version] [--config-dir] [--data-dir] [--runtime-dir]
               [--paths] [--json] [--debug]
               [subcommand]

Jupyter: Interactive Computing

positional arguments:
  subcommand     the subcommand to launch

optional arguments:
  -h, --help     show this help message and exit
  --version      show the versions of core jupyter packages and exit
  --config-dir   show Jupyter config dir
  --data-dir     show Jupyter data dir
  --runtime-dir  show Jupyter runtime dir
  --paths        show all Jupyter paths. Add --json for machine-readable
                 format.
  --json         output paths as machine-readable json
  --debug        output debug information about paths

Available subcommands: bundlerextension console contrib dejavu execute kernel
kernelspec migrate nbconvert nbextension nbextensions_configurator notebook
qtconsole run server serverextension troubleshoot trust

Jupyter command `jupyter-labextension` not found.

[martinRenou]

Thanks for opening an issue. Please keep in mind that the CSS is not considered as public API in ipympl, so if you do some hacks in the CSS they might not work from one version to the other.

There have been some changes on the Toolbar recently though. So you now have new properties that will change the visibility of the toolbar:

fig.canvas.toolbar_visible = 'fade-in-fade-out'  # Default: will only show the toolbar when the mouse is on top of the plot
fig.canvas.toolbar_visible = False # Always hidden
fig.canvas.toolbar_visible = True # Always visible

But you also can change its position:

fig.canvas.toolbar_position = 'left' # Default
fig.canvas.toolbar_position = 'right'
fig.canvas.toolbar_position = 'top'
fig.canvas.toolbar_position = 'bottom'

Now if you want the toolbar back on the top-left, I would suggest to set the toolbar_position property to 'top' and this CSS hack:

sthtml = """
<style>
.jupyter-matplotlib-toolbar {
    inset: 3px auto auto 3px;
}
</style>
"""

Also, if you want to display the plot between widgets. I'd suggest not using the Output widget, and instead, turn off the interactive mode:

plt.ioff()

Then use the canvas widget where you need it:

VBox((HTML(), fig.canvas, IntSlider()))

[martinRenou]

Concerning the last point, ipympl deserves a proper documentation for sure. I will try to find time to work on this.