Table footnotes

[darobin]

It is customary for table footnotes to be attached to the table. It is unclear how best to capture that semantically, though. There are various options, none really pleasant:

• Put them in the figcaption for the figure that contains the table. (That's what I've implemented so far, but it's not great.)
• Put them in the table's own caption — but that seem wrong, especially since that is already where we put the caption.
• Put them in a tfoot cell that colspans the width of the table. Sort of sensible, but the colspan bit feels hacky.
• Put them in an ol inside the figure, after the table. This is legit. Note that semantically that means that the figure is also representing the ol — but I think that's not incorrect. The ol would have the right DPUB ARIA markup of course.

I tend to lean towards the last option. Any thoughts?

[petermr]

Is there always only one table per figure? [Many publications have multiple tables, figures, etc. in the same container]. If so then I'd agree with [4] , else we have to have something attached to the table itself which would have to be [2] or [3] - in which case I'd favour [2].

[darobin]

This is something that is not described yet but that we need to cover: figure composition (#53). The theory is that every table is always contained in a figure, so we always have a single container available. That has many advantages in terms of styling and modelling, including the one you point at here.

Then if you can compose figures inside of another. So if you have two tables in a table figure they become 1A and 1B.