Re: encodings and cross references

    The current idea is that things like @ubaraccent{r} and the corresponding 
    encoded character, whatever the encoding, lead to the same representation,

I agree that this is logical.

... however ...

I have a big general concern about this.  In order to implement this, it
will be necessary to treat accented chars (@'e on up) differently in a
cross-reference node name than anywhere else.  In effect, we're
specifying a unicode backend just for cross-manual xrefs!  This does not
seem practical to me.

The complications are due to two things:
1) @-commands in node names.
2) 8-bit characters in node names.

And neither of these are currently supported (in general).

So it is seeming more and more premature to me to try to define a
"standard" specifying their usage, right down to which Unicode
(de)composition (or lack thereof) to use.  The fact is, these features
don't exist, so trying to guess the right thing to do with them seems
basically impossible to me.

So once again I'd like to suggest that we define the cross-manual xref
standard *without* @-commands and *without* 8-bit chars.  This will
cover 99.99% (at least) of existing Texinfo documents, and will be
useful right away.

We can deal with the complexities later, when there's actually support
for them in the language.  After all, it's a lot more important to
support 8-bit and accented chars *within* a manual than in cross-manual
xrefs!  I see no particular harm in delaying that aspect of things.
Sure, it would be nice if we could absolutely predefine the entire
standard and never change it again, but it doesn't seem to be working out.

Let's get what's useful now out there and implemented.  That's a big
enough cud to chew, for starters.

Thanks,
karl

P.S. As for your questions, in my reading about Unicode, I seem to
recall them stating that the decomposition into base characters +
accents is preferred, and the precomposed characters are there primarily
for compatibility and efficiency.  Base characters + accents will handle
the general cases.  I don't see any point in using the precomposed characters.
_______________________________________________
Texinfo home page: http://www.gnu.org/software/texinfo/
[email protected]
http://ff0.org/mailman/listinfo/texinfo-pretest