v3, SVG: \smash leads to 0 height

[pkra]

E.g., \smash {\mathbb{R}^2} renders to an SVG with height 0 (and similar viewport).

[dpvc]

It looks like 0-height SVGs are not displayed, as you probably noted. We should probably make a small minimum height (and width) to prevent that.

[pkra]

It looks like 0-height SVGs are not displayed, as you probably noted. We should probably make a small minimum height (and width) to prevent that.

I don't quite understand. Why would the SVG have height 0? I'd expect the output to appear identical to just \mathbb{R}^2 (as in CHTML).

[dpvc]

The height is 0 because you asked for the height to be 0 (that is what \smash does). The output will be just \mathbb{R}^2, but it has no height. That is also the case in CHTML. (At least the <mjx-math> element is height 0. The <mjx-container> ends up with a few pixels of height for reasons that I don't fully understand, but it is at least in part because it is an in-line element, and there is a minimum height for those that has something to do with the line-height and font metrics of the surrounding font.)

The CSS for the <svg> elements produced by MathJax v3 have overflow:visible, so even though the height is 0, it should still display the content (unless you have not included the MathJax CSS in your page). The problem is that a 0-height <svg> seems not to be rendered at all, so it needs a small amount of height for that to happen. That is why I said to use a minimum height.

[pkra]

The height is 0 because you asked for the height to be 0 (that is what \smash does).

Right, but I was thinking of this as "internally"

The CSS for the elements produced by MathJax v3 have overflow:visible, so even though the height is 0, it should still display the content (unless you have not included the MathJax CSS in your page).

Ah! I had not thought overflow works this way with SVG, especially not with the viewbox height. Thanks for explaining!

[dpvc]

I have made a PR to correct the problem of having a 0 height.

[pkra]

This persists for me (on the amermathsoc branch) on both Blink and WebKit based browsers (Gecko is fine).

Since the height was a very small ex value I tried manually changing it to 1px - this seemed to fix it. So it's possibly a rounding issue.

[pkra]

As a temporary workaround: svg {min-height: 1px!important;} seems sufficient.

[dpvc]

Thanks for the work-around. I think that may be the easiest way to handle it (using a selector that is specific to the MathJax output).