Markdown reader: use title of implicit figure as short caption.

[tarleb]

The title of an implicit figure, if set, is used as the short caption of
a figure. The short caption of a figure replaces the full caption in the
list of figures.

Closes: #7915

[jgm]

I think my main question is still the one raised in #7915 -- what if formatting or math is needed in the short caption?

One approach to this would be to parse the title as markdown.

[tarleb]

Another thought: The title could also be used as the image's alt text. This would be more consistent with the fact that titles cannot contain markup. I'd give accessibility higher priority than convenience for short captions, and this could be a good way to specify a description that differs from the figure caption. OTOH, it might be confusing to have implicit figures and normal images behave so differently.

[jgm]

Yes, I agree both that this makes sense and that it's probably too confusing to have implicit figures and normal images behave differently.

We can still specify alt text using an attribute, no?

![the caption](file.jpg){alt="My alt text"}

[jgm]

I suppose one alternative to overloading the title would be to allow a short caption to be specified via an attribute:

![the caption](file.jpg){shortcaption="My short caption"}

Drawback: English keyword. Drawback: no formatting.

[tarleb]

Am I right in thinking that pandoc currently only has special treatment for English keywords in the writers, but not the Markdown reader? It might be a dangerous precedent.

Would it make sense to support the table caption syntax on images?

![Close-up shot of the colorful face of a male mandrill.](mandrill.jpg)

: A male mandrill

Then the problem would become the more general issue of how to add short captions with that syntax. I guess the downside is that this is ambiguous syntax that could also be intended to mark a description list with an image term?

[jgm]

Hm, that does make a lot of sense. I agree that it's ambiguous, and that's a strike against it, but it is still tempting, and how common is it to use a definition list with an image term?

I would suggest that if we did this, then when this explicit caption is present, the image description should be the short caption, and the thing after : should be the long caption (since here we can include block-level content).

[tarleb]

If we could find a way to specify the short caption by extending the long caption syntax, then we could use that with tables as well. E.g., something like

![alt](a.png)

:short caption:
    long caption.

or, inspired by org-mode description lists,

![alt](a.png)

: short caption :: long caption.

Fuller example:

![The colorful face of a male mandrill.](mandrill.jpg)

: "Mandrill" test image :: The "Mandrill" standard test image, sometimes
  erroneously called "Baboon", is a popular sample photo used in image
  processing research.

Maybe this could also open a path towards more general, multi-image figures, but I don't know yet how that might look like.

[jgm]

: "Mandrill" test image :: The "Mandrill" standard test image, sometimes
  erroneously called "Baboon", is a popular sample photo used in image
  processing research.

That's not bad, actually!