For XML2RFC version 3 there is a new extended figure element, namely: it can contain multiple <artwork> or <sourcecode> (new in v3) elements. This in turn meant that mmark needed some way to signal “these artworks belong together”. Now Scholarly markdown has a syntax for this. It uses a special section syntax and all images in that section become subfigures of a larger figure. Disadvantage of this syntax is that it can not be used in lists. Hence mmark uses a quote like solution, just like asides (A>), but for figures: we prefix the entire paragraph with F> .

Basic usage:

F> {type="ascii-art"}
F>      +-----+
F>      | ART |
F>      +-----+
F>  Figure: This caption is ignored in v3, but used in v2.
F>
F>  ``` c
F>  printf("%s\n", "hello");
F>  ```
F>
Figure: Caption for both figures in v3 (in v2 this is ignored).

In v2 this is not supported so the above will result in one figure. Yes one, because the fenced (with the printf) code block does not have a caption, so it will not be wrapped in a figure.

To summerize in v2 the inner captions are used and the outer one is discarded, for v3 it is the other way around.

The figure from above will be rendered in v2 as:

                              +-----+
                              | ART |
                              +-----+

          This caption is ignored in v3, but used in v2.

                     printf("%s\n", "hello");

There is no v3 renderer (yet). Adding a second caption results in:

F> {type="ascii-art"}
F>      +-----+
F>      | ART |
F>      +-----+
F>  Figure: This caption is ignored in v3, but used in v2.
F>
F>  ``` c
F>  printf("%s\n", "hello");
F>  Figure: Another caption that is ignored in v3, but used in v2.
F>  ```
F>
Figure: Caption for both figures in v3 (in v2 this is ignored).

Becomes:

                              +-----+
                              | ART |
                              +-----+

          This caption is ignored in v3, but used in v2.

                     printf("%s\n", "hello");

      Another caption that is ignored in v3, but used in v2.