Ref #1100 , tickmode set to invalid value in mpltools

[nikhase]

• basevalidator throws an error on the boolean

Associated Issue: #1100

[jonmmease]

Thanks @nikhase, a few comments/thought below

[jonmmease]

I think it would be better to be explicit here and use 'auto' rather than None

[nikhase]

Can you maybe help me to understand what is going on? Are you trying to catch the axis scaling properties {“linear”, “log”, “symlog”, “logit”} from matplotlib.axes.Axes.set_xscale (other axis respectively)? There is no auto option. Maybe None is treated as auto?
It seems like despite the tickmode is set to None, it is inferred later on, for example the parsed JSON shows:

'layout': {'autosize': False,
               'height': 360,
               'hovermode': 'closest',
               'margin': {'b': 41, 'l': 61, 'pad': 0, 'r': 79, 't': 14},
               'showlegend': False,
               'width': 434,
               'xaxis':{...
                         'type': 'linear',
                         ...},
               'yaxis': {'anchor': 'x',
                         ...
                         'type': 'linear',
                         ...}
...

[nikhase]

btw, i chose None in the first place because of the code snippet axis_dict.pop('tickmode', None) in line 491.

[jonmmease]

Seems like this should be tick mode of 'log'. Would you mind trying this out on a simple matplotlib figure with a log axis?

[nikhase]

This is a matplotlib log example:

With None, it looks like:

and brings the warnings:

plotly.graph_objs.Annotation is deprecated. 
...
Converted non-base10 x-axis log scale to 'linear'
...
Dang! That path collection is out of this world. I totally don't know what to do with it yet! Plotly can only import path collections linked to 'data' coordinates

With log, it looks like:

Same warning as in the None setting here.
I cannot see a difference between None and setting to log. What should it then be?
Despite that, there are obviously troubles converting non base10 logs.

[jonmmease]

Thanks for looking into setting the axis type to values other than None. Based on the fact that the type is inferred to be linear later on, I think your solution of just changing False to None makes sense.

Thanks for the contribution!