On Mon, Apr 27, 2026 at 01:52:52AM -0600, [email protected] wrote: > Werner LEMBERG <[email protected]> wrote: > > > > > > I am wondering if it wouldn't be better to have non-alphabetic > > > characters at the end of the index instead of at the beginning. I'd > > > have to go see what books from O'Reilly and Pearson do to see what's > > > normal in the publishing world. > > > > The index of “The LaTeX Companion”, third edition (Addison-Wesley), > > which has a length of 150(!) pages, starts with 'Symbols', followed by > > 'Numbers', then 'A', 'B', etc. > > > > > > Werner > > So then we can leave well enough alone, with the symbols first.
What could be an improvement is to sort other non-ASCII UTF-8 initials after A-Z, not before. This would help the situation for Swedish, Norwegian, Danish and Icelandic which have extra characters at the end of the alphabet (probably other languages I missed). If we sort it by Unicode codepoint order, the characters will be out of order, but as long as they are after Z it will be an improvement. Correct order for Icelandic: ... X Y Ý Z Þ Æ Ö Codepoint order: ... X Y Z Æ Ö Þ Not the same, but still better than having them before A. (This would also require additions to the translation file txi-is.tex to output the proper UTF-8 sequences in the sort key for the three extra letters after Z.) When I checked on an online Icelandic-English dictionary, the accented characters like ý were sorted as accent variations of the base letters, so there would not need to be separate sections like "Ý" in between Y and Z, which (as far as I can tell) would be very hard to accomplish with the current setup. Likewise there is Đ between D and E, but this letter appears not to be used at the beginning of words, so not a problem for the index initials. Spanish has Ñ between N and O, although as far as I can tell this letter hardly occurs at the beginning of words. The best we can do is sort any the ñ... words after all the n... words, but still under the "N" heading.
