Re: Odd default schema behavior with SQL MQT's

[CRPence <CRPbottle@xxxxxxxxx>]

  Is "first library on the library list" meant to imply *CURLIB as 
the first in the user\current [CUR above USR] portion of the library 
list?  If not, then library QGPL should be the *CRTDFT for *SYSTEM 
naming with SQL, just as outside of the SQL.?

  Note that for unqualified CREATE of INDEX and VIEW in *SYS 
naming, the dependent object is not created into the *CURLIB [nor 
generally the first library in the user library list], but into the 
library of the first based-on [database file] object referenced as 
target of the SELECT or the library of the physical data for the 
access path.

  Since an MQT is not officially a dependent built over the data as 
with the VIEW and INDEX [i.e. it is physical data], QGPL as *CRTDFT 
seems like the proper effect for a CREATE when no *CURLIB.  That is, 
whatever CREATE TABLE would effect when not an MQT.  I would expect 
if the *CURLIB was established, then the claim that the TABLE is 
created into the "first library on the library list" [the CUR of 
*USRLIBL] would not differ between the MQT and the non-MQT CREATE 
TABLE.  If the *CRTDFT is in effect, then the MQT & TABLE should 
both go into QGPL [which need not be in *LIBL at all],  but a CREATE 
of a VIEW & INDEX could be into any library in *LIBL; i.e. where its 
[first] based-on file is located.

Regards, Chuck

David Gibbs wrote:
> I just observed some very odd behavior regarding SQL MQTs &
> DB2/400.
>
> I'm running the RUNSQLSTM command, with the DFTRDBCOL parameter
> set to *NONE, in an interactive job with a specific library list
> set.
>
> When I create a index, view, or table the PF or LF is created in
> the first library on the library list.
>
> If, however, I create a MQT with the same environment the MQT is
> created in QGPL instead of the first library on the library list.