- From: Ville Skyttä <ville.skytta@iki.fi>
- Date: Wed, 18 Feb 2009 01:25:35 +0200
- To: public-qa-dev@w3.org
- Cc: olivier Thereaux <ot@w3.org>
On Tuesday 17 February 2009, olivier Thereaux wrote: > Hi Ville, > > On 12-Feb-09, at 11:13 AM, Ville Skyttä wrote: > > Ok, I made links relative and bundled everything else except the > > donation > > program stuff in the tarball, and just uploaded 4.4 to CPAN. > > Great! It's on CPAN now indeed. I saw that the CpanTester seem to > point at a number of FAIL. > http://www.cpantesters.org/show/W3C-LinkChecker.html#W3C-LinkChecker-4.4 > > the error seems to consistently be: > “Can't locate LWP/RobotUA.pm” > > Odd... requiring LWP in Meta.YML should be enough to ensure that > RobotUA is functional, shouldn't it? Yep, I thought so... Practically all the FAIL notices come from the same tester. I'm already discussing this with him as well as other helpful folks on cpan-testers, and the problem is related to the -T (taint) option and the tester's environment. In his setup, LWP is not in the usual location but somewhere in $PERL5LIB, but use of -T means that $PERL5LIB is ignored. There are some ideas how to fix this, I hope it'll be done fairly soon. In addition to the above, there's one from a native Windows user, caused by lib/W3C/LinkChecker.pm's timestamp not being "up to date" wrt. its Makefile dependencies in the dist tarball which results in it being regenerated, but the Makefile target responsible for that apparently does not work in native Windows shell. I don't know how to make the target Windows shell compatible, but the problem has been worked around in CVS by trying to make sure that the file should be always up to date (and thus no regeneration needed) in future dist tarballs. So, nothing that serious, it seems. However due to the amount of FAIL reports against 4.4, I think a 4.5 release soon after the issues have been resolved would not be a bad idea to make the scary search.cpan.org failure count go away ;)
Received on Tuesday, 17 February 2009 23:26:12 UTC