On Wed, 2008-01-30 at 18:33 +0000, David Cantrell wrote: > On Wed, Jan 30, 2008 at 11:39:06AM -0500, Sam Tregar wrote: > > On Jan 30, 2008 9:22 AM, <[EMAIL PROTECTED]> wrote: > > > t/02yaml........# No tests run! > > > dubious > > > Test returned status 255 (wstat 65280, 0xff00) > > Amazing. Does every CPAN tester have a broken XML::SAX install? > > I don't have an XML::SAX install at all until your module lists it as a > pre-requisite, at which point I install it. > > > Is there > > something wrong with the way you guys are installing XML::SAX? > > Sort of. The fault appears to be with XML::SAX, which is broken > when installed using CPAN.pm.
That certainly didn't used to be the case and must have started happening quite recently. Have you any idea which versions of CPAN.pm are affected? Exactly this problem used to occur with older versions of CPANPLUS. The cause was that the XML::SAX makefile attempt to create the ParserDetails.ini file in the same lib directory that XML::SAX is in. The problem with CPANPLUS was that it included the blib directories in @INC during 'make install'. This behaviour was different to CPAN.pm which included blib in @INC during 'make test' but not 'make install'. Jos changed CPANPLUS to be consistent with CPAN.pm in release 0.77_03. If CPAN.pm has now changed this behaviour then I guess I'll have to come up with another way for the makefile to determine where XML/SAX.pm was installed (possibly along the lines you suggest). As I haven't seen the problem myself, any hint on affected version numbers would help me to test a fix. Regards Grant
