Re: texinfo.tex indexing bug

[Karl Berry]

    I have some old versions of BWK awk that can be used for testing.
    (I think as far back as ~ 1993; I'd have to check.)

I don't think it's important to go that far back.
My main concern would be working with some awk that standardly comes
with Solaris and BSD systems.

    To that end, can I assume that there will never be both \primary and
    \entry for the same key in a texindex input file?

There is no Texinfo document-level interface to \primary and \secondary,
as things stand, and they aren't used.  I hadn't actually realized there
was support for them in texindex before now.  chassell or rms wrote that
stuff however many decades ago, probably because they had some Big Idea
that never actually got completed, or maybe for some FSF printing of a
manual, and the Texinfo part never got back into the main line.

So, feel free to omit \primary and \secondary in any rewrite of texindex.
In fact, I'd go so far as to say it would be desirable to forget about them.

    It looks like texinfo.tex defines a @cindexsub command, but this is
    undocumented? How would it be used?

Yes, it is undocumented and shouldn't be used.
(I wouldn't know how to do so. :)

I feel like inventing more @FOOindex commands would be the wrong way to
go for handling sub(sub)entries and see(also).  My inchoate thoughts are
more about commands used on the index line, like
@cindex primary @subind{secondary}
or maybe the primary term needs to be marked too, like
@cindex @indmain{primary} @indsub{secondary} @indsubsub{tertiary}
Yeah, probably that's better.

Current
@cindex foo
would be equivalent, I think, to
@cindex @indmain{foo}

And for see/also, something like:
@cindex @indsee{something}
@cindex @indmain{something} @indseealso{something}

You can see why I feel this is not ready for prime time.
Not counting from the small matter of implementation :).

karl