> For example, the "General Index" in the Texinfo manual has sections
> for the following characters:
>
> ! " # & ' ( , - < > ? _ ` @ 8
>
> Each only has one or two index entries listed.
>
> I think it would be more typical to list all of these in a single
> section. I think listing them all under "*" would be better.
This might be typical for the Texinfo manual. However, for LilyPond,
to name an example, having multiple sections for non-alphanumeric
characters makes sense IMHO, see attached image. But maybe this is
beside your point :-)
> Running "pdftex equal-sort-keys.texi" produces:
>
> $ cat equal-sort-keys.cp
> @entry{*}{1}{@backslashchar {} (backslash)}
> @entry{*}{1}{/ (forward slash)}
> @entry{another index entry}{1}{another index entry}
>
> However, texindex produces the following:
>
> $ cat equal-sort-keys.cps
> @initial {*}
> @entry{@backslashchar {} (backslash)}{1}
> @initial {A}
> @entry{another index entry}{1}
>
> You can see that there actually only two index entries there. The
> two index entries with sort key "*" have been merged together.
>
> I remember that the convention of merging index entries with the
> same sort key was very old (from when I looked at this before,
> several years ago), but I thought we could reconsider this, as I do
> not actually see any advantage of merging the entries.
>
> Does anyone have an opinion on this?
I support the reconsideration to not merge them (probably as a new
default, with an option to revert to the previous behaviour).
> I also thought that not merging index entries would work well when
> entries differed only by accents (or maybe even only by letter
> case). For example, if an index file had:
>
> @entry{Bogen}{1}{Bogen}
> @entry{Bogen}{1}{Bögen}
>
> - texindex could compare the actual text of the entries (which are
> different, as "o" is different from "ö") when sorting the index
> entries, not just look at the sort keys. Then the index sorting
> would be close to perfect for several European languages.
Sounds good to me.
Werner
