Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: nlssort

Re: nlssort

From: Peter Gulutzan <pgulutzan_at_ocelot.ca>
Date: 18 Nov 2002 09:04:28 -0800
Message-ID: <36c478c6.0211180904.1d326807@posting.google.com>


"Howard J. Rogers" <howardjr2000_at_yahoo.com.au> wrote in message news:<g%RB9.78558$g9.221344_at_newsfeeds.bigpond.com>...
> "Peter Gulutzan" <pgulutzan_at_ocelot.ca> wrote in message
> news:36c478c6.0211170850.66b97193_at_posting.google.com...
> > "Howard J. Rogers" <howardjr2000_at_yahoo.com.au> wrote in message
> news:<G8wB9.77968$g9.219593_at_newsfeeds.bigpond.com>...
> > > Not that I've ever felt the need to use it myself, but if you are not
> happy
> > > with Oracle's own sorts, you can build your own. There's a utility
> provided
> > > in (I think) 8.1.6 and above that does just this... Locale Builder.
> > >
> > > Might be worth checking out.
> > >
> >
> > Hegyvari Krisztian and flatline are unsurprisingly correct about the
> > rules -- I based my erroneous remark on an old Hungarian-English
> > dictionary, but now I see that Á always follows A according to current
> > standards, including the Posix spec.
> >
> > In that case, Oracle8 should be doing what it says it does --
> > Hungarian sort order -- and if it does not (which I am not claiming, I
> > only accept flatline's statements) is it the responsibility of "not
> > happy" Hungarians to work around? I agree that one can devise one's
> > own routine, or compare one character at a time with substring
> > functions, etc. The questionw, though, were: is this a bug, and does
> > anything similar happen in other countries which uses Latin-2
> > characters (Croatia, The Czech Republic, Poland, Romania, Slovakia, or
> > Slovenia)?
> >
>
>
> I wasn't talking about functions doing character comparisons or building
> your own routines. There is a utility, called Locale Builder, that lets you
> build your own NLS libraries to have the system sort in any order you care
> for. You need only then set the NLS environment variables, and not a single
> piece of code would be required.
>
> I don't think its particularly reasonsable to criticise Oracle for "not
> doing what it says its doing... Hungarian sort order" when your own post
> mentions "current standards" -implying that there are -or might be-
> (possibly old and obsolete) other standards to which it may be complying.
>
> It's just the same with French as she is spoken in Paris sorting in subtly
> different ways to French as she is spoken in, say, Cameroon.
>
> The tool is there to let you be happy. Unhappiness would be being told to
> accept what Oracle provides or lump it.
>
> Regards
> HJR
>
> > Peter Gulutzan

The words "build your own" and "not happy" were yours, and so is the word "criticise" -- I tried to avoid criticism and I see none in the original poster's question. There are a few people who would be interested in knowing which standard (yes there are choices) Oracle followed for linguistic sorts. In the unlikely event that it has/had deviated from whichever standard it has chosen, then the original poster has reason to enquire whether there is/was a bug, as he did. In the event that Cameroon has a different standard than France, I would be interested in knowing about that too.

I must have been unclear in my original message, and so I apologize.

Peter Gulutzan Received on Mon Nov 18 2002 - 11:04:28 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US