Actually, Nigel, could you please nopaste/gist the whole CPAN config file? (Removing anything sensitive of course.)
Thanks On Jul 4, 2016 5:03 PM, "Andreas Koenig" < andreas.koenig.7os6v...@franz.ak.mind.de> wrote: > >>>>> On Mon, 04 Jul 2016 09:29:21 -0400, Nigel Horne <n...@bandsman.co.uk> > said: > > >> This one has also delivered the following suspicious tests: > >> > >> | > http://www.cpantesters.org/cpan/report/10bca32a-3386-11e6-9fc5-14bfbd15da07 > >> | | Eval-Closure-0.14 | 2016-06-16 | > >> | > http://www.cpantesters.org/cpan/report/5145d9c6-3aec-11e6-9fc5-14bfbd15da07 > >> | | Class-Unload-0.09 | 2016-06-25 | > >> | > http://www.cpantesters.org/cpan/report/4cc4ad60-3b04-11e6-9fc5-14bfbd15da07 > >> | | Reply-0.40 | 2016-06-25 | > >> | > http://www.cpantesters.org/cpan/report/fe856154-402a-11e6-9fc5-14bfbd15da07 > >> | | Perl-PrereqScanner-1.023 | 2016-07-02 | > >> > >> Can you reproduce the fails? Does this ring any bell? Maybe if we can > >> identify the reason on this smoker, we have it resolved for all others. > > > Yes, I can reproduce it, when I run it by hand. Even though CPAN does > > say "Warning: prerequisite Test::Fatal 0 not found.", it doesn't > > actually download and install for the build process. > > Thanks for trying. May I ask that you reveal another config variable, > `trust_test_report_history`? > > The line you quote above ("Warning: prerequisite Test::Fatal 0 not > found.") usually comes from MakeMaker. CPAN.pm should neither see it > nor write it (please correct me if I'm wrong). When CPAN.pm tells > something about a prereq it should look like > > ---- BEROV/DBIx-Simple-Class-1.009.tar.gz ---- > DBD::SQLite [build_requires] > DBIx::Simple [requires] > > At least with current CPAN.pm. > > Look, here is the pass report of this smoker for Test-Fatal-0.14: > http://www.cpantesters.org/cpan/report/16d71e5e-1039-11e6-8057-de3b705340c6 > > It was a success. So if the configuration sets trust_test_report_history > to true, it would not try to build Test-Fatal again. That would probably > explain the whole issue. Otherwise, we will have to reiterate again > other options... > > -- > andreas >
