Re: cross manual references in html manuals, second

[Karl Berry]

    It also seems to me that @verbatim, @verb, @tex and @html should not be
    allowed too, but I am not certain about these, however (especially @verb
    and @html).

@verbatim, @tex, and @html are environments (with corresponding @end),
so they can't be included in node names.

@verb, on the other hand, seems like it could be allowed.
It would just "expand" to its argument.

    Accented letters are transformed into their 8-bit equivalent character,
    according to the iso latin 1 mapping.

There are accented letters not in Latin 1.  Perhaps we have use utf-8.

    The following @ commands have no real reason to be used in node names, thus
    it is recommended to use the plain text equivalent:

I'm not sure what this means.  I don't think it matters though.

    @equiv equiv OR ==
    ...

I think the symbolic representation (==) would be better than the word
`equiv'.

    @today today

Should probably be the actual date.

    @copyright (C)
    ...
    @exclamdown ! OR 8-bits equivalent
    @questiondown ? OR 8-bits equivalent
    @pounds pounds OR 8-bits equivalent

The 8-bit character seems like the right thing for all of these.

    I don't remember what the following does ;-)
    @tie SPACE

It produces an interword space at which a line break is not allowed.
Like ~ in plain TeX and LaTeX.

    @email is replaced by the text, and if not present the mail adress.

@email could appear in a node name as much as anything else.

    The letters in hexadecimal should be in small caps.

Lowercase, not small caps.

    (the local software may also skip the file name as browsers (or servers ?)

Servers.  Browsers just do `GET /' or whatever.  It's up to the server
to decide whether to return index.html, or index.htm, or run index.cgi,
or whatever.  (See the DirectoryIndex directive in Apache.)

    The manual name should only contain the following characters:
    [A-Za-z0-9-_/], / having a special meaning.

And `.', also with a special meaning.  (For example, the Emacs manuals
say address@hidden ../info/emacs'; not that I think this is a good idea,
but it's the way it is.)

    and anchor names and generate a file per filename constructed as above. 

Really?  I don't think we necessarily have to generate a file for every
node.  (If that's the `filename constructed as above' that you mean.)
That would make --no-split impossible to implement, and users wouldn't
be happy about that.

Instead, how about always generating the xref to the top level of the manual?
http://distant/manual/index.html#target

In the no-split case, that will just work.  If the distant manual is
split (in whatever way), then the distant software makes the above url
redirect (using your javascript method) to:
http://distant/manual/targetfile.html#target

    Also makeinfo keeps the caps in file names.
    
Did I miss something?  Don't we get to use both A-Z and a-z?  Are you
proposing case-folding somewhere along the way?  I didn't see that.


Thanks,
karl