Re: proposal for html manual cross links

> Perhaps we should require that the user specifies the charsets
> somehow, like from the command line, and refuse to generate output if
> she doesn't?

I am not sure this will solve the issue if the charsets for the target and the
source of the link are different. But maybe I wasn't clear on what was 
problematic to me. 

I see a problem if there are 2 manuals. For this example, I suppose that 
e acute isn't in encoding2 or hasn't the same encoding than in encoding1.

file target.texi

   @documentencoding encoding1
   @node Top
   @top

   @node node with an e acute encoded in encoding1 é
   @chapter
   @bye

file link.texi. The issue is what to put instead of ???

   @documentencoding encoding2

   @node Top
   @xref{node with an e acute encoded in encoding1 ???,,,target}.

I don't see any possible way to do cross references with different encodings.

> Sorry, I'm confused: you specifically say in your proposal (or at
> least I think you did) that there are situations where some nodes will
> be unreachable.  AFAIK, the current code in makeinfo never leaves
> nodes that are not reachable.  If that's not so, please show an
> example, and let's discuss it.

Here is an example:

@node @b{truc} @u{A} @point{}
@node <b>truc</b> A( -!-

both lead to <a name="%3cb%3etruc%3c%2fb%3e%20A(%20-!-">
I attached a manual with that code.

In fact the proposal sometimes disambiguate less than the existing code. 
Indeed with the proposal,

@node {@b{a node},,,manual}
@node {@i{a node},,,manual}

are not disambiguated although they are with the current code. 

The real issue here is what nodes should we consider as equal for the 
purpose of cross references (the same questions would be interesting in a 
more general context, that is when are nodes considered to be equal).

Maybe people reading this list could try to answer the following question.
Should the following pair of node names be considered as equal (for the 
purpose of html manual cross references and/or in documents):

a  node
   a   node

@'u
@'{u}

@TeX{} text
@TeX{ text}

@b{my node}
my node

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

@dotless{i}
i

@point{}
-!-

@u{A}
A(

@u{B}
B(

The 4.6 code finds an error for the first pair only. 
With --html and with nodes this code shows 
yes yes yes no no (I believe there is a bug) yes yes yes yes

With anchors the result is very different as @-commands are not 
expanded when generating the anchor (which leads to incorrect
cross references...).

With the proposal the responses would be
yes yes yes yes yes yes no no yes

And you, what do you think about these node pairs ?

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