On Tue, 15 Nov 2016 18:37:41 -0800 Bart wrote: BVA> Apparently that build failure has been fixed in commit "better BVA> handling of pdu max message size; now per packet, not per BVA> session". However, since that commit all Perl regression tests BVA> fail: BVA> BVA> make[1]: Entering directory 'perl/SNMP' BVA> PERL_DL_NONLAZY=1 "/usr/bin/perl" "-MExtUtils::Command::MM" BVA> "-MTest::Harness" "-e" "undef *Test::Harness::Switches; BVA> test_harness(0, '../blib/lib', '../blib/arch')" t/*.t BVA> Can't call method "get" on an undefined value at t/async.t BVA> line 45. t/async.t ......... BVA> Dubious, test returned 1 (wstat 256, 0x100) BVA> Failed 20/20 subtests BVA> # Failed test 1 in t/bulkwalk.t at line 42 BVA> # t/bulkwalk.t line 42 is: ok(defined($s1)); BVA> Can't call method "bulkwalk" on an undefined value at BVA> t/bulkwalk.t line 53. t/bulkwalk.t ...... BVA> Dubious, test returned 1 (wstat 256, 0x100) BVA> Failed 62/62 subtests BVA> t/conf.t .......... ok BVA> Can't call method "get" on an undefined value at BVA> t/dump-session.t line 27. t/dump-session.t .. BVA> Dubious, test returned 1 (wstat 256, 0x100) BVA> Failed 1/1 subtests BVA> [ ... ]
How are you running configure? I ran configure with defaults plus --enable-perl-modules and did not have any test failures. I did peek and perl/SNMP/SNMP.xs, though, and I'm guessing that the issue is local netsnmp_sessions being created with snmp_sess_init() being called (or snd/rcv max sizes not being set). Robert ------------------------------------------------------------------------------ _______________________________________________ Net-snmp-coders mailing list Net-snmp-coders@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/net-snmp-coders
