On Mon, Oct 12, 2026 at 05:22:34PM +0200, Patrice Dumas wrote:
> On Mon, Oct 12, 2026 at 06:02:20PM +0300, Eli Zaretskii wrote:
> > Something like having the node written as (assuming French
> > translation):
> >
> > Node: L'indice^?[Index]^?
> >
> > where ^? are the DEL characters (or some other special sequence).
>
> The issue with such possibilities is that they are not backward
> compatible, past Info reader cannot render correctly the node name.
>
> It could still be a better idea, especially for the middle term than a
> new tag in the tag table, as what you propose would not lead any Info
> reader to malfunction, there would only be a weird node name. A new tag
> could still be the best long-term solution, but it could only be used
> years away from 2025.
It is close to what I was going to suggest: add an anchor to the tag
table giving the location of an index. The anchor name wouldn't usually
be visible to users.
I know you referenced a mail where you opposed checking for "Index" in
an anchor name, but I also didn't go into any detail on that occasion (I
can't remember what I was thinking at the time).
It could look something like:
Ref: Index cp^??894490
It could reference the exact byte of where the output from @printindex
started, rather than just the node.
This has the advantage of also recording the name of the index from the
Texinfo source ("cp" in this example for the "concept index").
(We'd have to have appropriate checks that this didn't clash with user
defined node or anchor names, etc.).
The use of the English word "Index" in this context would be part of the
the Info format, similar to other words such as "Node" or "Next". It does
not mean that we are showing that word to users - the word "index" could
be something completely different in a different language.
If we thought a clash with a user-defined target name was likely we could
use something more distinctive, like "Texi Index cp" maybe using control
characters as Eli suggested.
