xsl-list
[Top] [All Lists]

RE: Natural Sort

2006-01-25 02:00:00
Thank you, substring-after and substring-before are perfect.

The real world data actually has the 3 types of domain names

EG:

<server dn="media1.mysite.com" id="server392" os="FreeBSD 5.1"/>
<server dn="media2.mysite.com" id="Server488" os="Fedora Core 1"/>
   ...
<server dn="media10.mysite.com" id="Server433" os="Fedora Core 1"/>

<server dn="members1.mysite.com" id="server322" os="Debian Linux"/>
<server dn="members2.mysite.com" id="Server435" os="Fedora 
Core 2"/>
   ...
<server dn="members10.mysite.com" id="Server491" 
os="FreeBSD 4.11"/>


The internationalisation commmunity (I18N) have spent about twenty years
talking about the subtleties of different collating sequences used in
different countries, and this has resulted in large libraries of collations
available in many different environments, so if you want to collate the way
that was popular in Sweden in the 19th century, you can almost certainly do
so. But one of the things that really irritates me about this work is that
it has been so slow to recognize the kind of alphanumeric sorting you
describe here. There's a brief mention in the Unicode Collation Algorithm
[1], but it's explicitly excluded from the algorithm proper. I don't think
there's any support for it in the Java collation mechanisms, despite all
their power and complexity. 

In a back-of-book index you want "ISO 8859" to appear before "ISO 10646",
and it's usually very hard to achieve. However, in the last year or two I've
noticed that an increasing number of products do in fact use this "natural"
sequence, for example when listing the files in a directory, Windows XP
explorer lists test9  before test10. So perhaps the world is slowly waking
up.

I'm going to think about how I can incorporate such a feature into Saxon's
collation mechanisms, despite the lack of Java support.

[1] http://www.unicode.org/reports/tr10/ (Search for "numeric order",
section 1.4, and see 1.9.2, "non-goals", point 2)

Michael Kay
http://www.saxonica.com/



--~------------------------------------------------------------------
XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list
To unsubscribe, go to: http://lists.mulberrytech.com/xsl-list/
or e-mail: <mailto:xsl-list-unsubscribe(_at_)lists(_dot_)mulberrytech(_dot_)com>
--~--



<Prev in Thread] Current Thread [Next in Thread>