On Jan 7, 2008 6:07 PM, David Cantrell <[EMAIL PROTECTED]> wrote:
> Usually when they grumble about particular distributions being tested
> it's because a tester has gone and tested something hideously ancient.

That's why I think that telling them to go update META.yml isn't an
answer.  (Though it should be an option as well.)


> A danger, however, is that if we were to all suddenly start skipping
> everything by author FOOBAR then the normal way to do that would be to
> just not download his modules.  That means that anything that depends on
> his modules would also not get tested.  Which is bad.  It would be nice
> to come up with a mechanism that would only affect the sending of test
> reports - so then testers would still download my stuff if something
> else depends on it, but just not bug FOOBAR with failure reports.  If
> CPAN::Reporter and YACSmoke could share the same 'blacklist' file that
> would be awesome.

That's a good point.  I'll probably work up some config options for
CPAN::Reporter to specify an exclude list.  I'm trying to get a
virtual machine smoking rig going at home -- doing so will be
motivation to extend/fix CPAN::Reporter in helpful ways like that.

> YAML, BTW, while a great format for us humans to update, is probably not
> the best one to use while actually running tests.  To read it requires
> the YAML module, and I like to test on a bare-bones perl, which doesn't
> have YAML.  But I can easily turn YAML into perl code using
> Data::Dumper, so please please please support that as well, like Andreas
> does for distribution prefs files.

How about something even easier -- like MANIFEST.SKIP format: pure
text, one regex per line checked against the "CPAN-style" distribution
name.  (I.e. "DAGOLDEN/CPAN-Reporter-1.0601.tar.gz")

Thus:

# TESTERS.SKIP
^CRANKY
^UNCLE
^\w+/Acme
^\w+/Ancient-Distro-Shouldnt-Be-Tested

On reflection, I think that would be a nice example of the KISS
principle in action.

I'll look to add a "skipfile" option to the next CPAN::Reporter to
jump start this, then we can figure out a repository to use.

David

Reply via email to