🚀 go-pugleaf

Here are more consolidated answers to the original question from Luis
Miguel (luism@sybase.com):

     What are the most commonly sort orders (collating sequences) used
     for the (economically) major non-English European languages?

     Spanish

     German

     French


Jeff
INSOFT-L List Manager

*========================================================================*
 Jeff Bowyer                         EMail:
 Computing Center		     jbowyer@cis.vutbr.cz
 Technical University of Brno
 Udolni 19, 602 00 BRNO
 Czech Republic
*========================================================================*


Sender: Glenn Adams <glenn@metis.com>
        Technical Director, Unicode Consortium

  Patrick O'Callaghan <poc@usb.ve> wrote:
    >  German
    >
    >  There are two different collating sequences in wide spread use.
    >
    >  Classically, you simply ignored the umlaut, sorting a" as a, etc.
    >  The s-tset (looks like a Greek beta) sorts as 'ss'.

It is not sufficient to merely sort some element x as another element y.
The reason this is insufficient is that such an algorithm does not produce
reliable results; that is, given an arbitrary input, in may produce an
arbitrary output order (precisely because it is treating some element x
as another element y).  A stable sort algorithm must produce a fixed
output order independent of its input order.  Therefore, it is necessary
to specify orderings of elements that, in linguistic terms, are considered
to have equal weight.

  Patrick O'Callaghan <poc@usb.ve> wrote:
    >  French
    >
    >  Just ignore the accents (and the cedilla), and sort as in English.

This algorithm is entirely inadequate, not only for the reasons I mention
above, but also because it doesn't account for the fact that specific
orders regarding the resolution of homographs with different accents are
in fact used in French lexicography.

Some excellent papers on this subject include:

"Quand <<Z>> vient-il avant <<a>>?" and "R\egles du classement alphab/etique
en langue franc,aise et proc/edure informatis/ee pour le tri" written
by Alain LaBont/e, at the Minist\ere des Communications, Gouvernement du
Qu/ebec.


Sender: John X. Laporta <@gte.com:johnl@krell>
        Senior Member of Technical Staff
        GTE Laboratories Incorporated

  Patrick O'Callaghan <poc@usb.ve> wrote:
    >  French
    >
    >  Just ignore the accents (and the cedilla), and sort as in English.

Libraries use this since accents on capital letters are
optional. Otherwise one might have two works by the same
author that would alphabetize to different places in a
list of works sorted by author.

But there is also a sorting algorithm for French that takes
equivalence classes into account. The Canadian standard describes
it, and this is used as an example in Posix 1003.2. I could
forward this if there is interest.

Collisions are less frequent than expected because the consistent
orthography of French makes certain combinations of (accented)
vowels and consonants impossible. One would not write "pe`r"
for example.


Sender: Keld Simonsen (keld@dkuug.dk)

There is work underway in international standards committees
on this subject.

ISO/IEC JCT1/SC22/WG15 POSIX is producing a standard ISO/IEC 9945-2:1993
POSIX shell and utilities which has provisions for collation
specification together a number of other internationalization/localization
items. The standard has just been approved and it is undergoing
minor editorial editing before publishing as an international standard.
It is expected to be published later this year.

WG15 is also collecting POSIX locales and making them publically
avaiable by email and anon ftp from the site dkuug.dk in the directory
i18n . About 30 locales are available covering major European languages
like German, French, Spanish, Italian, English, Dutch, Portuguese,
Hungarian, Chech, Danish, Swedish, Norwegian, Finnish, Estonian,
Latvian, Lithuanian, Polish, Greek, Russian etc.

To get the info by email you can mail mail-server@dkuug.dk
help on its use is obtained by having in the body of the message:

       begin
       help
       end

Many of these locales are coming from official national ISO
standardisation organisations, being in the process of formal
standardisation in the respective countries; but some are just
private contributions.

Most of them are defined in a symbolic character set independent way
on a large subset of ISO/IEC 10646-1:1993, and applicable to
a number of coded character sets. Most of the locales
are defined with multi-level collation, for example for letters
first ignoring accents and case, then ignoring only case and eventually
discriminating between upper and lower case.

Special cases like the Spanish ch and ll, or the Danish/Norwegian ae
letter are also handled in the respecitive specifications.

The European Standardisation Organisation CEN is also working
on collating issue and is in the process of producing a European
character string collating standard, for use in CEC (Commission of
the European Communities) institutions and other places.
CEN is also producing a registration standard for cultural elements
including POSIX locales, and is in the process of collecting
information from the CEN members and associated members. Many of the
POSIX locales in the WG15 collection originates from this effort.

I am the editor of the two CEN standards mentioned and also the
keeper of the WG15 locale collection. You are welcome to put
further questions on the locales to me - then I will forward them to
the relevant persons.


Sender: Emmanuel Uren <76107.405@CompuServe.COM>
        Power Up Software

I also referred Luis Miguel to the January 1992 issue of Windows/DOSz
Developer's Journal that focusses on Internationalization.  There's an
article by Peter Gulutzan "How to Compare Characters with European Collating
Sequences".  And Denis Garneau's "Keys to Sort and Search for Culturally