Remove the scaleanchor constrain on axes having an image trace

[josephernest]

In Keep square pixels + ability to have any rectangular shape zoom with Plotly.js, with a plot with 2 layers's answer by @lvlte, we see that it is sometimes useful to be able to remove the scaleanchor constrain applied by default on axes having an image trace.

But this doesn't work: setting scaleanchor: false on either or both axes has no effect.

The easiest solution is to patch, in https://github.com/plotly/plotly.js/blob/v2.25.2/src/plots/cartesian/constraints.js#L34,

hasImage: axHasImage[axName]

by

hasImage: false

but this is not a solution in the long-term.

Do you think it could be useful to add a setting to avoid this manual patching?

[alexcjohnson]

Good point! I don't see any way currently to turn this behavior off. Seems like the options are:

• add something like false to the acceptable options for scaleanchor so you can explicitly override a default set by having an image
• add an attribute to image traces to disable automatic scaleanchor - ie giving the image trace an attribute like squarepixels which would default to true but you can set it to false in which case that image would not contribute to axHasImage.

The second might be more discoverable in the documentation, but it's less direct and if you have multiple images you'd need to set this on all of them. The first requires you to figure out that the constraint is specified on the Y axis and that an explicit false will remove it, but you only have to do this once (per Y axis) and would generalize if we ever add any other conditions that automatically set a constraint.

So I'd lean toward the first, plus a note somewhere in the docs for image trace about this setting in the Y axis.

[josephernest]

This would be great @alexcjohnson! What do you think @lvlte?

[lvlte]

Yes I would go for the first option too. I think I will make a PR this weekend or next week.

[josephernest]

Thanks for the pull request @lvlte and for the merge!
Do you think it will be included for v2.25.3 @archmoj?

[archmoj]

Thanks for the pull request @lvlte and for the merge! Do you think it will be included for v2.25.3 @archmoj?
@josephernest New features would be included in the next minor.
We would possibly release 2.26.0 this week.