Re: Index advisor advising index on primary key

Bottom line:

System value QSRTSEQ is *LANGIDSHR which is then propagated through job descriptions and the RUNSQLSTM command creating the table. The table ends up with a SRTSEQ of QSYS/QLA10025S. (upper/lower case considered equal). This is what I want.

The default collating sequence in Qry/400 is hexadecimal (*HEX). This is what's causing this index (and other indices) to be advised.

That can be changed interactively query-by-query (read "labor intensive").

What I'll be looking for:

1) A way to change the default for Qry/400 so new queries default the way I want.
2) A way to get a list of queries that have the *HEX collating sequence attribute (like some DSPOBJD to a *OUTFILE kind of thing).
3) A way to programmatically change the attribute of existing queries (I'm not too hopeful on this).

If anyone already knows how to do any of the above 3 things, please point me to the command. In the meantime, I'll be searching . . .

Jeff Crosby wrote:

Initial reply from IBM on this is: it's the SRTSEQ attribute of the table.

System value QSRTSEQ is *LANGIDSHR which is then propagated through job descriptions and the RUNSQLSTM command creating the table. The table ends up with a SRTSEQ of QSYS/QLA10025S. (upper/lower case considered equal).

I asked some questions, so stay tuned for more info . . .

Jeff Crosby wrote:

I opened a PMR.

Here's an interesting tidbit. Just because, I did the following:

CRTDUPOBJ OBJ(DMITMMST) FROMLIB(DBMSTF) OBJTYPE(*FILE)
TOLIB(JEFF) CST(*NO) TRG(*NO)

Note that I said CST(*NO). A DSPFD on the newly created file says:

Number of constraints . . . . : 0

but further down the DSPFD says:

Access Path Description Access path maintenance . . . : MAINT *IMMED Unique key values required . : UNIQUE Yes Access path journaled . . . . : No Access path . . . . . . . . . : Keyed Constraint Type . . . . . . . : NONE Number of key fields . . . . : 1 Record format . . . . . . . . : MBRECORD
Key field . . . . . . . . . : ITNBR Sequence . . . . . . . . : Ascending
Sign specified . . . . . : UNSIGNED
Zone/digit specified . . : *NONE Alternative collating seq : Yes So there's no 'constraint' but there's still a unique access path. Furthermore I *was* able to do an ALTER TABLE and *add* a unique constraint on ITNBR in this duplicated file.

Are we having fun yet?
rob@xxxxxxxxx wrote:

Keep me posted. One attendee at the session said he built the recommended index and it still wanted him to build the recommended index. I'm curious if that is the case, or the unique constraint, or ...

By the way, the person who really shines at these sessions is an IBMer by the name of Tom McKinley. He's the "John Sears" of our time.

Rob Berendt