Eli Zaretskii wrote:
> FWIW, during the development of Texinfo 4.0, a good-faith effort was
> made to do as you say in 1).
I appreciate that, thank you.
> The large number of different browsers
> notwithstanding (each one with its unique set of bugs ;-), we actually
> tested the produced HTML with as many as we could get our hands on. I
> don't know if X-Mosaic was one of them, though.
That's what I am for:-)
> > Another point, about the <dl> variant: one side benefit would be that
> > you could provide links back from the footnote to the footnotemark;
>
> IIRC, <dl> didn't work well (I think we used <dl> originally, but
> switched to <ol> for some good reason).
Ok, I had a look at the code to see what the problem was. It's not
from the structure of the code, so I assume there was a problem with
some browser; unfortunately there's no comment talking about it.
However, I found that it's quite easy to fix the empty anchor problem
in the following way:
<li>
<p><a name="fn-40">We use ... </a></p>
Here's a patch for that (apply in directory makeinfo):
--- footnote.c~ Mon Sep 20 14:20:52 1999
+++ footnote.c Wed Jun 7 21:14:45 2000
@@ -326,10 +326,10 @@
if (html)
{
/* Make the text of every footnote begin a separate paragraph. */
- add_word_args ("<li><a name=\"fn-%d\"></a>\n<p>",
+ add_word_args ("<li>\n<p><a name=\"fn-%d\">",
footnote->number);
execute_string ("%s", footnote->note);
- add_word ("</p>\n");
+ add_word ("</a></p>\n");
}
else
{
AFAIK the only shortcoming of this method is: if there is something in
the footnote that is not allowed to be in an anchor (e.g., a paragraph
break), the resulting HTML is not standard. However, I guess this
case is rare, and I doubt any browser will have problems with it.
- anton
