Ingres 9.2 - Why does Ingres think these are the same?

#21

On Jan 24, 6:03*am, "James K. Lowden" <jklow... (AT) schemamania (DOT) org>
wrote:

Quote:

On Mon, 23 Jan 2012 01:21:09 -0800 (PST)

Kristoff <kristoff.pic... (AT) ingres (DOT) com> wrote:
On Jan 21, 10:36*pm, "James K. Lowden" <jklow... (AT) schemamania (DOT) org
wrote:
the allkeys file, the Default Unicode Collation Element Table,
which says:

$ grep -E '^00.+(LOW LINE|SPACE)$' allkeys.txt *| head
0020 *; [*020A.0020.0002.0020] # SPACE
005F *; [*021B.0020.0002.005F] # LOW LINE
...
I'm not saying this is what Ingres is doing. *I'm saying this is a
valid interpretation of what I understand to be the controlling
standard.

I'm not speaking about the actual default collation at unicode.org,
but the old one which came with the 2.0 standard. Take a look at:
http://unicode.org/Public/UCA/2.1.9/basekeys-2.1.9.txt

$egrep '^005F|^0020' basekeys-2.1.9.txt

0020 ; [*0209.0020.0002.0020] # SPACE
005F ; [*0209.0021.0002.005F] # LOW LINE; COMPATSEQ

So in this old collation sequence the first weight of both chars is
the same.

Thanks for the clarification, Kristoff. *If I may, did the standard
also change how/when the second field comes into play? *AIUI
the current standard, each level is a tie-breaker for the the one
before. I won't go into detail now in case it's irrelevant.

To the OP's question, is the Unicode collation user modfiable (without a
compiler) or contemplated for a future release?

Regards,

--jkl

Yes, the collation can be changed - there is no "true" collation which
would fit into any language, so actually it is expected that user will
customize it.
For Ingres you need to compile the collation sequence, but the
compiler (aducompile) is delivered with the Ingres distribution.

Be careful when creating your own sequence or when using the very
latest available at unicode.org. Currently Ingres does not understand
all the actual unicode features. If you look at the udefault5
collation which is now delivered in addition to the old one, it is not
the original one coming with Unicode 5.0, it only use the weights
defined in that version, but only contains those characters defined in
standard one coming wit Ingres.

Kristoff

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Ingres 9.2 - Why does Ingres think these are the same?

comp.databases.ingres comp.databases.ingres

Re: Ingres 9.2 - Why does Ingres think these are the same? - 01-24-2012 , 03:24 AM