Andreas J. Koenig wrote:
Evidence is rather strong to me that CPANPLUS::Dist::Build 0.44 is
producing bad reports.
Several of my statistics indicate that CPANPLUS::Dist::Build=0.44 never
produces PASS reports and instead produces many false FAIL reports.
Here a few samples:
http://pause.perl.org:3000/reports_by_field?distv=ack-1.92;field=mod%3ACPANPLUS%3A%3ADist%3A%3ABuild
http://pause.perl.org:3000/solved?distv=Business-OnlinePayment-SagePay-0.10
http://pause.perl.org:3000/reports_by_field?SUBMIT_xxx=Submit&distv=Devel-NYTProf-2.11_949&field=mod%3ACPANPLUS%3A%3ADist%3A%3ABuild&field=meta%3Afrom
http://pause.perl.org:3000/reports_by_field?SUBMIT_xxx=Submit&distv=POE-Component-Cron-0.021&field=meta%3Afrom&field=mod%3ACPANPLUS%3A%3ADist%3A%3ABuild
http://pause.perl.org:3000/reports_by_field?SUBMIT_xxx=Submit&distv=TAP-Formatter-JUnit-0.03&field=mod%3ACPANPLUS&field=mod%3ACPANPLUS%3A%3ADist%3A%3ABuild&field=meta%3Afrom&field=mod%3ACPANPLUS%3A%3ADist%3A%3ABuild
It seems that I have identified about 40 distros that each got more than
10 false FAIL reports.
Please turn off your smokers if you're using CP:D:B 0.44
Let me know if you need more data.
Thanks,
Hello,
Thanks for warning us about this! I quickly looked at some of the
reports and I can tell you with confidence that a lot of my FAIL reports
is bad. I don't think it's CP:D:B's fault but the VM's fault. It seems
like their local time is skewed/slowed/whatever and it causes confusion
when running the makefile. Here's one sample report:
http://www.nntp.perl.org/group/perl.cpan.testers/2009/12/msg6478612.html
I've now turned off the smoker on those VMs and is investigating the
time problem. I'm using NetBSD 5.0.1 and FreeBSD 7.2-RELEASE on a
VirtualBox 3.0.12 host running Ubuntu 9.10. If anyone have any tips,
please let me know!
P.S. Great use of your new regression analyzer! Big kudos to you :)
