No matter what I have tried so far, I get a core dump in test 55 of plsql.t.
I'm on Solaris 2.5.1 with perl 5.6.1, DBI 1.15, and Oracle 7.2.2. I've
tried a plain make, and a make with the suggested -b, -c, -l, and -n
LIBCLNTSH, all result in:
t/plsql.............dubious
Test returned status 0 (wstat 139, 0x8b)
test program seems to have generated a core
DIED. FAILED tests 55-63
Building with -g doesn't seem to help with gdb on the core:
Program terminated with signal 11, Segmentation fault.
"/usr/platform/SUNW,Ultra-1/lib/libc_psr.so.1": not in executable format:
File format not recognized.
#0 0xef554f04 in _end ()
Building with -p causes it to fail worse:
t/general...........install_driver(Oracle) failed: Can't load
'blib/arch/auto/DBD/Oracle/Oracle.so' for module DBD::Oracle: ld.so.1:
/opt/perlnew/bin/perl: fatal: libclntsh.so.1.0: can't open file: errno=2 at
/opt/perlnew/lib/5.6.1/sun4-solaris-2.5.1/DynaLoader.pm line 206.
at (eval 1) line 3
Compilation failed in require at (eval 1) line 3.
Perhaps a required shared library or dll isn't installed where expected
The core dump happens at the last line of this part of plsql.t:
my $cur_query = q{
SELECT object_name, owner FROM all_objects
WHERE object_name LIKE :p1 and ROWNUM <= 3
};
my $cur1 = 42;
#$dbh->trace(4);
my $parent = $dbh->prepare(qq{
BEGIN OPEN :cur1 FOR $cur_query; END;
});
ok(0, $parent);
ok(0, $parent->bind_param(":p1", "V%"));
ok(0, $parent->bind_param_inout(":cur1", \$cur1, 0, { ora_type =>
ORA_RSET } ));
ok(0, $parent->execute());
BTW, I don't even understand that SQL, so maybe I can get by without it
working :-)
I've stepped through the test in the debugger and everything seems okay
until the call to _setup_handle:
696==> DBI::_setup_handle($h, $imp_class, $parent, $imp_data);
697
698: return $h unless wantarray;
699: ($h, $i);
700 }
701 # minimum constructors for the tie's (alias to XS version)
702: sub DBI::st::TIEHASH { bless $_[1] => $_[0] };
At this point, a single step takes me to Carp::longmess with a @_ of
'SIGSEGV'. If instead I look at $h, I see this:
DB<3> x $h
0 DBI::st=HASH(0x5862dc)
SV = PVMG(0x586df8) at 0x587c40
REFCNT = 1
FLAGS = (ROK)
IV = 0
NV = 0
RV = 0x532dc4
dbih_getcom handle 'DBI::st=HASH(0x532dc4)' is not a DBI handle (has no
magic) at /opt/perlnew/lib/5.6.1/dumpvar.pl line 154.
SV = PVMG(0x586df8) at 0x587c40
REFCNT = 1
FLAGS = (ROK)
IV = 0
NV = 0
RV = 0x532dc4
SV = RV(0x3e1828) at 0xefffeed8
REFCNT = 1
FLAGS = (ROK,READONLY)
RV = 0x532dc4
Issuing rollback() for database handle being DESTROY'd without explicit
disconnect() at /opt/perlnew/lib/5.6.1/perl5db.pl line 1483.
Debugged program terminated.
Well, that can't be right.
FWIW, I've been using these Oracle libraries successfully on Solaris 2.5.1
with perl 5.004_04 and a DBI of 0.89 and DBD::Oracle of 0.46 (don't laugh).
Any ideas? This is the only test that fails - well, I don't know what the
remaining 7 subtests would do, obviously.
Yeah, yeah, 7.2.2 is old. My next step is to essay 8.1.6 on Solaris
2.6. But I still need something working on Solaris 2.5.1, and Oracle 8.1.6
is not supported there.
--
Peter Scott
Pacific Systems Design Technologies
http://www.perldebugged.com
