Re: Spanish sorting works properly only for locale "e" and not for "es_ES"

[Andy H. <an...@jt...>]

<Zdenek.Macura@Sun.COM> writes
> I have found an issue with the sorting for Spanish locale:
>
> We can read on
> http://oss.software.ibm.com/icu/userguide/Collate_Concepts.html
>
> that character sequence ch is considered as a single letter, but I have
written
> a programme and I see that it doesn't work.

'ch' is treated as a single letter in the locale es_TRADITIONAL.
It is not treated that way in es_ES.

You can see the rules for tailoring the sort order for different locales.
Look at the file icu/data/es_TRADITIONAL.txt, which includes the line
    "&c < ch <<< Ch <<< CH"

This line is not present in es.txt or es_ES.txt
The precise meaning of these collation tailoring rules is a bit on the
obscure side, but this is mostly an issue when writing them.  It's pretty
easy to check whether special handling for particular characters have been
introduced into a locale or family of locales.

(please send any follow up messages to ic...@os..., rather than
icu4c-support, which is intended primarily for build and porting issues.)

   -- Andy Heninger
      hen...@us...