Bug #125279 for Win32-OLE: Use of uninitialized value in OLE.xs Variant::Type() function

Hi, I downloaded the module from CPAN and compiled by myself. The 'make test' produces some warning messages when performing test on t/2_variant.t: % make test "/usr/bin/perl.exe" -MExtUtils::Command::MM -e 'cp_nonempty' -- OLE.bs blib/arch/auto/Win32/OLE/OLE.bs 644 PERL_DL_NONLAZY=1 "/usr/bin/perl.exe" "-MExtUtils::Command::MM" "-MTest::Harness" "-e" "undef *Test::Harness::Switches; test_harness(0, 'blib/lib', 'blib/arch')" t/*.t t/1_nls.t ....... ok t/2_variant.t ... 1/38 Win32::OLE(0.1712) error 0xffffccb0 at t/2_variant.t line 39. eval {...} called at t/2_variant.t line 39 Win32::OLE(0.1712) error 0xffffccb0 at t/2_variant.t line 40. eval {...} called at t/2_variant.t line 40 Win32::OLE(0.1712) error 0xffffccb0 at t/2_variant.t line 73. eval {...} called at t/2_variant.t line 73 Win32::OLE(0.1712) error 0xffffccb0 at t/2_variant.t line 94. eval {...} called at t/2_variant.t line 94 : I found the warning messages come from an issue in OLE.xs function Type(). There is a flow control according to variable ix and then a CheckOleError() is called with variable hr. IT seems hr value is uninitialized on this call if the flow went through ix=0,3,4,5. A patch to fix it is as attached.