On Tue, May 05, 2015 at 07:42:14AM -0600, J.D. Laub wrote:
> I'd like to propose a small enhancement to DBD::Oracle.
> However, when I connect via DBI using an oracle wallet (SEPS) entry, that
> information is never
> presented. I tracked the inconsistency to
> DBD-Oracle-1.74/lib/DBD/Oracle.pm , line 309:
>
> 309 unless (length $user_only) {
> 310 $user_only = $dbh->selectrow_array(q{
> 311 SELECT SYS_CONTEXT('userenv','session_user') FROM
> DUAL
> 312 })||'';
> 313 $dbh_inner->{Username} = $user_only;
> 314 # these two are just for backwards compatibility
> 315 $dbh_inner->{USER} = $dbh_inner->{CURRENT_USER} = uc
> $user_only;
> 316 }
That 'uc' looks suspicious to me, but that would be a pre-existing issue.
> I'm thinking maybe it should save off the existing values for those 3
> attributes, run the query, and
> then append the old values onto those resulting from the selectrow_array.
> (Technically that would leave
> them in the wrong order - with the selectrow_array info ahead of the
> connect info - but maybe that's not
> important.) The attached patch rather clumsily does that, and has
> resolved the problem in my
> environment. I'm happy to revise the patch - just tell me where it's
> lacking.
> unless (length $user_only) {
> + # It may be we've already encountered a warning by this point,
> + # such as "ORA-28002: the password will expire within %d days".
> + # We'll cache it for reinstatement.
> + my ($err, $errstr, $state) =
> + ($dbh->err, $dbh->errstr, $dbh->state);
> $user_only = $dbh->selectrow_array(q{
> SELECT SYS_CONTEXT('userenv','session_user') FROM DUAL
> })||'';
> + # Now we'll reinstate the earlier warning. We're just
> + # appending it, so in the extremeley unlikely case that the
> + # selectrow_array we just issued also issued a warning, the
> + # 2 warnings will appear out of order.
> + $dbh->set_err($err, $errstr, $state) if defined $err;
Looks ok to me.
Tim.
