On Sun, Aug 9, 2009 at 10:05 PM, wrote:
> Hi CPAN testers,
Hi, Craig.
> and perhaps fix the first 2 as well. I have the following
> questions which I hope you will have time to answer:
Usually someone will find some time to answer questions. Sometimes it
might take a day or two.
> 1) If I just change the tests, should I increase the version
> number and resubmit the whole package?
Yes. Every release to CPAN needs to have a new version number. (This
is a restriction impose by PAUSE -- the upload server). If only the
tests are changing, you might consider a "minor version" such as
0.5401. This is a convention that signals to people that changes are
minor.
> 2) If not, and I just re-upload the tests, will the relevant
> CPAN testers automatically learn of the updated tests?
CPAN Testers works off whatever is uploaded to CPAN. So when you
submit your package to PAUSE, CPAN Testers will start testing it,
usually within a few hours. However -- some test reports come from
individual users, not automated smoke testers. So you may not see
particular platforms in your test reports until a user runs tests
manually. (E.g. attempting to install your module again).
> 2a) If 1=no and 2=yes, is it OK to email the specific testers
> who had the failures and ask them to rerun the tests as
> soon as they can?
>
> 3) If failures persist (particularly on the first 2) is it
> possible for me to get debugging access to the system(s) in
> question? Or is this a Q better asked of the particular
> tester(s)?
We encourage CPAN authors to contact testers directly, either to
request they re-run tests or to ask for help debugging particularly
challenging issues.
Finding author emails is a little complex to keep spammers from
harvesting email addresses. You'll need the "NNTP ID" of the report,
which is the "ID" column on the cpantesters page for your module.
Then you plug the ID into this form:
http://stats.cpantesters.org/cpanmail.html
As a side note, I see that you have a lot of "UNKNOWN" reports from
testers -- almost certainly test machines without the mm library. You
might want to check out Devel::CheckLib as a tool for your Makefile.PL
to check for the library and exit with a helpful error message when
the Makefile.PL is run.
Regards,
David