Re: ORDER BY and alternate collating sequences

Hi Joe

The LANGIDSHR gets you a sort sequence that is a lot more like some kind of ASCII than EBCDIC - I've not pursued it much more than that - we don't use ALTSEQ in any of the places it could be set, and it still works in USA settings.

Give it a try without any alternate sequence - have values that are numeric and alpha and punctuation - see what you get - it might just be enough to get the ASCII-like result you need.

I'm still using DDS to create the indexes, unless they are EVI. There's a SRTSEQ parameter there for setting *LANGIDSHR. I know, I know - but I have reasons!!

Just as JDBC has a setting, so does ODBC somewhere - either in a DSN setup or in a connection-string parameter.

On 3/27/2012 9:40 AM, Joe Pluta wrote:

On 3/27/2012 9:17 AM, Charles Wilt wrote:

On Tue, Mar 27, 2012 at 10:05 AM, Joe Pluta<joepluta@xxxxxxxxxxxxxxxxx> wrote:

Okay, bear with me, I'm old and slow<grin>.

What are the steps? Do you set the SQL session parameter to LANGIDSHR
in STRSQL and then run the CREATE INDEX command? Will that then create
an INDEX that is permanently LANGIDSHR?

Yes

So then can I end my session, bring up a new session (without
LANGIDSHR), use my index and it's magically still in LANGIDSHR
sequence?

No, as there's no way for you to explicitly "use the index"

If you configure your new session for LANGIDSHR, when you do a select
and specifiy ORDER BY indexedColumn, the system will implicitly use
the LANGIDSHR index you created.

Also, that only seems to address case sensitivity. What
tells it to use the alternate collating sequence that corresponds to
CCSID 819, for example?

Same applies AFAIK, but I've never tried it.

Okay, that removes the "magic black box" aspect which always makes me
uncomfortable. This makes much more sense. Having an index (or not) is
really not the issue; if the job is set for LANGIDSHR, the SQL engine
will use such an index if it exists. The problem is to get an ODBC job
to use LANGIDSHR. Or CCSID for that matter.

To set LANGIDSHR on a JDBC connection I can use "sort weight=-shared"
but I'm unsure of how to set the language or CCSID to get ASCII
collating sequence.

Joe