alphabetic sorting

alphabetic sorting

[Johannes Hüsing]

Dear Hans, dear list,

in German, as well as in other languages, there are different sorting patterns 
concerning the umlauts and the eszet.

There is a dictionary order, which sorts the letter "a (sorry I have only
a terminal available that wouldn't print all latin1 letters) in the same
position as a, in conflicting positions after it (so Bar comes befor B"ar).

The other one is the phone book order, which treats "a as ae, so J"ager and
Jaeger will be sorted at the same position.

Both orders make sense. In German, declination of many words comes with the
replacement of a vowel by its umlaut, as in jagen -- J"ager. In a dictiona-
ry you want to have the words Jagdrevier and J"agerlatein close together,
as they are related.

In names, you will have alternative spellings of the same word, and it may be
awkward to have to remember if somebody's name spells with an umlaut or with
the replacing vowel-e construction.

I found a good coverage of this peculiarity in 
http://www.rostra.dk/alphabet/alpha_dt.htm (in German only).

In a serious document management software you would like to have both types
of orders available, depending on whether you have a subject or a name index.
Hans, you told me that alternate order switches could be accomplished within
ConTeXt, if only you are told about these kind of things.

Gruss

Johannes

Re: alphabetic sorting

[Tobias Burnus]

Hallo Johannes, Hans and hi list members,

> There is a dictionary order, which sorts the letter "a (sorry I have only
> a terminal available that wouldn't print all latin1 letters) in the same
> position as a, in conflicting positions after it (so Bar comes befor B"ar).
"a == ä

> The other one is the phone book order, which treats "a as ae, so J"ager and
> Jaeger will be sorted at the same position.

Well using texutil --quote one gets "a -> a etc. (I think also "s->s
which is wrong):
#D \startopsomming[opelkaar]
#D \som  \type{\name}: csnames are stripped
#D \som  \type{{}}: are removed
#D \som  \type{\"e}: and alike are translated into \type{"e} etc.
#D \som  \type{"e}: is translated into an \type{e} and \type{b} etc.
#D \som  \type{||}: becomes \type{-}
#D \som  \type{\-}: also becomes \type{-}
#D \stopopsomming

I think we should define either sorting as default (probably ä->ae) and
support the other via overwriteing the sortkey (--quote is probably
applied latter so it would be obsolete).

The question is: Shouldn't \SS (=\sz, \SZ (?), "s, "S, "Z, "z) -> ss
(type b), æ = ae {type ?}, \oe -> oe. Þ (thorn) -> ?, ã, õ etc. be
treated "generically" for all languages so one only needs to overwrite
the values? I think ó, ò and ô are sorted as "o" in all languages.

Tobias

PS: Have a look at sortkey, l.493 in
texmf/tex/context/base/enco-ini.tex.