Re: proposal for html manual cross links

    I thought that this list would be better for such things. I can post on 
    bug-texinfo too if you wish.

texinfo-pretest is good for random development/feature discussions.

    text. For example '@TeX{a text}' leads to 'TeX a text'. 'SPACE' means

This should produce an error.  Argh, I see that it doesn't.  Ok, your
proposal describes current practice, if we make it an error in the
future, that'll be tough :).

    ---    corresponding utf-8 characters (hyphen) 2010
    --     corresponding utf-8 characters (em dash) 2014

I don't see the logic to transforming -- to emdash (should be endash if
anything), and --- to three hyphen chars.  Or is that not what you're
suggesting?

For purposes of xrefs, I think it might be simplest to just leave -
characters as themselves, untransformed (or transformed into 002d).
However many -'s are in the input are in the output.  Wdyt?

The utf8 equivalences mostly look good to me.  Some comments, though:

    @minus       utf8 character 2212 (or 002D ?)
2212, definitely.  @minus{} explicitly means the minus sign.

    @point       utf8 character 2217
@point{} isn't a math operator, just a random symbol, so I think 2605
(black star) might be better.  I don't see any "logical" character to be
used.  (Hey, maybe we would work 2602 umbrella or 2603 snowman into
Texinfo somehow?  They're so cute. :)

    ``     "
    ''     "
How about transforming each ` into 2018 (left single quotation mark) and
each ' into 2019 (right single quotation mark)?

    For '--' in normal text, it seems to me that the utf-8 character is '2010' 
    (hyphen), and for '---' it is '2014' (em dash).

In normal text, -- would be 2013 (endash).  (--- is 2014 emdash, yes,
and just - would be 002d.)  But in the context of node name
equivalences, see my suggestions above.

In this context, I'm not sure we buy anything with multicharacter
transformations according to what the text "really" means, as if we were
actually trying to typeset it.  The simplest possible
character-by-character interpretation seems the least prone to error?

    process all node and anchor names such that
    all the url leading to index.html#target

Sorry to be unsure on this, but does this mean that to follow these
links, either (a) the output must be monolithic (everything in one html
file), or (b) javascript is required?  Is there any other way to implement?


Finally, in general, I'm afraid I'm unclear about something fundamental:
could you discuss a bit about the relationship between utf8 and URI
escaping and how this will all work?  Is %XX hex representation used at
some level (and if not, why not?)?  Could you show some concrete
examples?


Thanks again for all your work on this.
karl
_______________________________________________
Texinfo home page: http://www.gnu.org/software/texinfo/
[email protected]
http://ff0.org/mailman/listinfo/texinfo-pretest