Limitations of Current Approach

• Requires deeply committed CPAN testers, both to test modules and to maintain infrastructure

• Dependent on CPANtesters.org's operational status

  • Reports sometimes not available for days after filing

• Difficult to search CPANtesters.org for FAILs

  • Inherently non-comprehensive

• No easy way in rt.perl.org to get list of all distributions failing due to a given commit

  • Have to scroll through many posts to a BBC ticket

  • Must manually record when those distributions once again PASS

• No way to measure progress over time

  • How does the number of currently "broken" distributions compare to the same time in the previous development cycle?

  • Limits Perl 5 committers' ability to anticipate CPAN "breakage"

  • Limits Perl 5 Porters' ability to hold Perl 5 committers accountable for CPAN "breakage"