On Fri, 04 Jul 2014 00:58:07 +0200, Barbie <bar...@missbarbell.co.uk>
wrote:
I hear about emails being ignored only for authors who get daily emails
with tons of PASSes and an occasional fail.
There are only 59 authors who get sent PASS reports, and all would
changed their preferences. As mentioned previously, the default is FAIL
only.
I'm not prepared to remove it, as some authors have previously asked for
it. I can only assume the authors you have spoken to are part of the 59
(60 including BooK) above.
Did the default not change in the past?
Or are you saying the database tracked explicitly who had changed their
settings from the default?
If not, it's still possible that these authors changed it in the past and
forgot about having done so. There's no other explanation for what
happened to BooK if the default didn't change.
I think even if these options are left, it'll be worthwhile to send them a
reminder and explanation email.
While the footers on every email report do mention the prefs page, it
doesn't specifically say that the frequency or type of reports can be
adjusted -- so adding that information may help.
I'll change that over the weekend. Thanks, saves me thinking up copy ;)
Maybe consider making it the header too? I'm worried that people who
generally ignore those emails won't ever notice the footer changing, if
they even look at the emails at all at any time.
What would be the consensus for the default? "FAIL,UNKNOWN" or
"FAIL,NA,UNKNOWN" ?
I would vote for all three, since ALL of them are failures, just
from different parts of the process.
NA reports are generally useless - usually they indicate that the
smoker is
running an incompatible OS or perl version. I have no interest in
receiving lots of NA reports for Moose on perl 5.6, for example.
(Is there an example of a *useful* NA report that deserves a
notification?
Personally I would agree. My emails only send me FAILs and wouldn't want
to change that. If I want to find out about others I use the website.
I'm not convinced it needs changing. I would rather send them just FAILs,
than get further flack for send more that most authors would be happy to
ignore. If they want the extra report information, they can adjust their
preferences.
Ah, i was under the impression NA indicates any crash of the build
process, not just a crash with output indicating incompatibility. Then
it's not useful to include as default.
Still, please do consider UNKNOWN for inclusion though, as it does,
according to CPAN::Reporter documentation, indicate a real failure case.
--
With regards,
Christian Walde
