Re: proposal for html manual cross links

> How about if we use the hex format _00XX of the 8-bit code, as if
> these characters were ASCII punctuation?

This seems the way to go, though the cross refs may be different if the
character is encoded differently in different 8-bit code, but maybe this is
not a general case.

> Is this example interesting?  We do discourage @-commands in node
> names; your examples not only uses them, but also abuses that to
> deliberately create node names which are literally different, but lead
> to the same text after expansion.  Texinfo is quite loose in some
> constructs, so it's easy to abuse it like that.  But are these
> examples practical?  That is, would anyone in their right mind ever
> use such names?

I agree this example is not interesting, it is illustrative. But my
thought was that if the proposal is to be used for cross manual references
then it should never (if possible) change. Thus I thought that it would be 
right to have a proposal which take into account all the theoretical issues 
we can imagine.

> They should be considered identical, but, with the current
> implementation of makeinfo, it is very hard to do so, since command
> expansion in node names imposes a tremendous performance hit, as
> demonstrated by the use of --commands-in-node-names.  If we do
> consider them identical, makeinfo will treat such nodes as an error
> and refuse to process the file.

Even 
@b{my node}
my node
should be considered identical ?

> Did you try --commands-in-node-names?

Impressive. With --commands-in-node-names 
makeinfo --html
finds that these nodes are the same, except for

@b{my node}
my node

and 
@^{@dotless{i}}
@^i

but I believe the second is a bug because the html is wrong (I will try on 
a box I have at home which can compile the cvs version).

And makeinfo doing info find all these nodes identical.

Hum. I am not certain about what it means for the cross manual references.
My first thought would be to consider that the ascii equivalent representation
should be used, in order to disambiguate as far as possible. But I am not
sure this is right, because in the first proposal I made I proposed a 
choice between the ascii representation and the unicode equivalent and 
Karl said that the unicode would be right (for example equiv was mapped to 
== or the unicode character). Maybe it revealed the fact that he think
that @equiv{} and == are fundamentaly different. What do you think about
that Karl ?

Anyway at least we can be sure that there remains only 3 cases which are, 
in my opinion not settled:

things like
   @equiv
   ==

like
   @u{A}
   A(

and
   my node
   @b{my node}

I have no idea about what is the right thing. I will put in the proposal
whatever is chosen.

Pat
_______________________________________________
Texinfo home page: http://www.gnu.org/software/texinfo/
[email protected]
http://ff0.org/mailman/listinfo/texinfo-pretest