Problem

I would like to plot scatter points as contextual data while preserving the current axes data limits. Something like this can be achieved withAxes.set_autoscale(False), but I would like to keep auto-scaling on, and disable it for a particular feature. As far as I understand this is how collections work, whichAxes.scatter uses in the background.

Context: this feature will allow GeoPandas to consistently plot line, polygon (usingadd_collection()) and point data (usingscatter()) as contextual data not affecting axes limits (seegeopandas/geopandas#2602). Down the line it will benefit a small library I am developing (hyoga), to plot gridded data with xarray, and geographic data for context with geopandas.

Proposed solution

I noticed thatadd_collection has anautolim keyword that defaults toTrue. As opposed to completely disabling auto-scaling, settingautolim=False simply skips updating the axes data limits for this particular collection, meaning auto-scaling stays on, but ignores it (if I read this correctly). I propose to propagate theautolim keyword argument toAxes.scatter and pass it toadd_collection here:

This would allow something like:

ax.scatter(x_subject,y_subject)# default to autolim=True, update limitsax.scatter(x_context,y_context,autolim=False)# don't update limitsax.plot(something_else)# update limits disregarding {x,y}_context

I am happy to open a PR if this makes sense.