Front page | perl.perl5.porters |
Postings from May 2009
On Sat, 30 May 2009, David Golden wrote: > On Thu, May 28, 2009 at 7:54 PM, George Greer <greerga@ein.m-l.org> wrote: >> If anybody has done the above, suggestions (and scripts) welcome. I'll >> attach what I have written so far to give you an idea of what I was trying. >> It doesn't work yet, due to lack of time, so don't run it. > > All my CPAN Testers smoking scripts (based on C::R::Smoker) are in a > git repository: > > git://echo.dagolden.com/git/cpan-testers-tools/ > > Browseable here: > > http://echo.dagolden.com/git/?p=cpan-testers-tools.git;a=tree > > All of it is undocumented and some of it uses various modules you may > need to install, but it's what I use to drive my smoker. > > You might want to look at regression.pl as it's what I was using to > regression test Module::Build. Also loop-smoke.pl takes a list of > perls and does a one-shot through each one before looping back through > all the perls again. Thanks, that'll give me something to refer to. My goal is slightly different though: I'm building Perl from Git branches (blead, maint-5.10, and maint-5.8) then trying to smoke CPAN against those. I'm more interested in what changed in Perl to break CPAN than what changed in CPAN to break CPAN, although I'll get quite a bit of the latter too. I'm not entirely sure CPAN Testers differentiates between 5.10.0~release and 5.10.0~maint-5.10 though. I did, however, finally get my script working, so I can do three (or more) simultaneous CPAN smoke builds on the different branches. Now I just need to prove the way I handled the Git repository keeps each branch updated and all I'll have left is the usual distroprefs work. I'm still taking suggestions for virtual machines to run too, like chromatic was writing about. If I don't hear anything I'll set up Windows 2000 with the same 3-tree smoke build as my Linux machine and that's good enough for me. -George GreerThread Previous | Thread Next