index vs. index_ vs. index- vs. index_0078

[Karl Berry]

Patrice and all,

(Background for the record.)

Simon Josefsson and others have pointed out that a node named "index"
ends up in the same index.html output file as the top-level HTML
output (the toc, etc.).  See the bottom of 
http://www.gnu.org/software/gnulib/manual/html_node/index.html as an
example.

We've bandied about various ways to deal with it: write the "index" node
as index_.html, index-.html, or index_0078.html.

I think index_ is preferable to index-, because index- could be parsed
as having a meaning now, namely a node name of "index " -- of course we
don't allow trailing whitespace in node names, but that happens at a
different level.  index_, on the other hand, is currently syntactically
invalid, given our use of _ to introduce Unicode characters.  Therefore
it can't be confused with anything other than the special case it is.

And I think index_ is preferable to Bruno's proposed inde_0078 for two
reasons. (1) As above, inde_0078 has a meaning now, namely "index".
(2) As I've mentioned in passing before, we shouldn't hardwire the word
"index", since occasionally sites use something else for their index
file.  That rule is slightly more complicated -- Unicode-ize the
last character of the name, vs. "append _".  I don't see any benefit to it.

So my conclusion is that our best approach is to say a node named "index"
should be referred to as "index_" in the HTML output.

Am I missing arguments or counter-arguments or the point entirely?

This is all assuming case-sensitive filesystems.  As we know, on
case-insensitive filesystems, we already have special cases to write an
"Index" node onto index.html as well.  I am not sure if it is worth
doing anything about this, or if anything needs doing.  I wouldn't like
to make another special case out of it.

Thanks,
Karl