xenops-increase-size changes forgotten after reveal

[JoeyEremondi]

Currently, I can use xenops-increase-size to make the size of the rendered TeX bigger, and it works quite well. The problem is, if I enter a snippet and reveal it to reveal the math command, when it renders again it goes back to the default size, not the increased size. I have to do "xenops-increase-size" followed by "xenops-decrease-size" to restore all the snippets back to their usual size.

Is this a known bug? Is something just configured wrong for revealing?

[dandavison]

Hi @JoeyEremondi. As a meta comment -- I'd love to do more work on Xenops but recently I haven't been writing any maths (I made it for myself when I was taking time out to study) -- and so it has been a bit hard to justify the time.

If I remember correctly, this was a known bug. Basically here

xenops/lisp/xenops-math.el

Lines 356 to 366 in 4d75c1c

	(defun xenops-math-image-change-size (element factor)
	"Change ELEMENT image size by multiplicative FACTOR."
	(-if-let* ((image (xenops-element-get-image element)))
	(when (eq (image-property image :type) 'svg)
	(image-flush image)
	(let* ((data (or (eval (image-property image :data))
	(and (f-exists? (image-property image :file))
	(prog1 (f-read-text (image-property image :file))
	(setf (image-property image :file) nil))))))
	(if data
	(setf (image-property image :data) (xenops-util-svg-resize data factor)))))))

this function takes in the factor by which to change size, but I think what was needed was some sort of global/default factor which would be honored on each render automatically; and then the xenops-increase-size function would mutate the global factor variable, or something along those lines. So this would give a uniform multiplicative display scaling of all images over and above whatever the default size would be based on the latex source.

If you have the slightest temptation to look into fixing it, that would be fantastic, and I can offer you my claim that the code is fairly clean and that it's fun to do xenops development, and obviously that I'd be happy to discuss it.

[dandavison]

Or perhaps the problem is the image caching? That when we change size, we need to invalidate the entries in emacs's image cache so that they are re-rendered with the new size.