> I'll change it to maybe a single column and add headings for
> starting letters, so they show up as navigation in the right column.
> This should also simplify the markdown file for the index.
Thanks in advance.
> In the API docs, there isn't any non-latin characters and accents,
> so using the Unicode collation algorithm completely (implementing or
> as a library) might be too much but
Yep.
> I will implement the basic sorting rules.
>
> I'm not completely sure why "role < Role < rôle" but I don't think
> this is important for docwriter.
The standard (US?) English sorting rules mandate that accented letters
are sorted after non-accented ones. Uppercase vs. lowercase is not as
strong, this you get the shown collation. However, it's indeed not
important for docwriter since C doesn't allow such identifiers.
The sorting on the index page should make sense now.