Yes. JL's detection that the formatting difference was due to different default column formatting because of the difference between the literal length and the length of one of the choices for a bind variable in support of the changed setting for cursor sharing was spot on and a great pick up. It also served as a great reminder that the mechanism of cursor sharing is factoring literals into a usefully spread possible set of bind variable lengths. IF you need a different column formatting, injecting (either in (g)login.sql or a script setting an individual sqlplus sessions depending on the scope you need) a statement similar to the one I suggested you test containing the type and heading you actually want should keep it from varying and give you an opportunity to supply the logical name for the column heading in place of a value based column name. mwf -----Original Message----- From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of David Yu Sent: Wednesday, March 26, 2014 10:31 PM To: mwf@xxxxxxxx Cc: <oracle-l@xxxxxxxxxxxxx> Subject: Re: output of a select statement Thanks Mark, I use the same sqlplus,actually,the databases running on the same server with the same oracle home . You must have seen Jonathan's comment on this, that's explain everything. Rgds David 在 2014-3-26,下午11:38,"Mark W. Farnham" <mwf@xxxxxxxx> 写道: > Dear Mr. Fishy, > > Did you in fact find a login.sql or glogin.sql difference, and did > > column "'a'" format a3 hea duh > > in your sqlplus session before the query in fact demonstrate that it > was a column formatting problem? > > (And I'm not sure I understand your mention of any cursor sharing bugs > in this context, or what was strange about it.) > > r, > mwf > > -----Original Message----- > From: ORACLE-L [mailto:mr.fishyu@xxxxxxxxx] > Sent: Wednesday, March 26, 2014 9:27 AM > To: fuzzy.graybeard@xxxxxxxxx; mwf@xxxxxxxx > Cc: oracle-l@xxxxxxxxxxxxx > Subject: Re: output of a select statement > > Thank you for your reply,there's another issue within > 10.2.0.3,cursor_sharing set to similar or force will trigger the > bug.after setting to exact,this issue fixed too.a bit strange :) > > Rgds > > David > > 在 2014-3-26,上午12:00,Hans Forbrich <fuzzy.graybeard@xxxxxxxxx> 写道: > >> On 25/03/2014 7:30 AM, ORACLE-L wrote: >>> Hi Group, >>> >>> I have two DBs10.2.0.4 running on Solaris 10. >>> >>> And the output on these two databases are totally different from sqlplus: >> Are these run from the same client? Are they from a client on each > machine or perhaps from a separate client? What happens if you use > the same client to access each DB? >> >> Two things I would check to start >> >> 1) Any login.sql/glogin.sql >> 2) Language/Territory related settings at the client and on the >> Solaris > machines. >> >> /Hans >> -- >> //www.freelists.org/webpage/oracle-l > > -- > //www.freelists.org/webpage/oracle-l > > -- //www.freelists.org/webpage/oracle-l -- //www.freelists.org/webpage/oracle-l