Today I'm pleased to announce the launch of the CPAN Testers' "Collated Email Notification of Tester Reports for Authors Service" [1], a centralised notification system to send authors email updates for the reports, that were previously sent by testers to the authors directly. David has now updated Test-Reporter to suppress the sending of mails to authors, and the CPAN Tester community are currently upgrading their environments with the latest version. As such the current direct mails should now tail off, with the new notification system taking over.
Following some heated discussions last week, part of the plans for CPAN Testers 2.0 have been brought forward. As much of the data recording and presentation is already fairly centralised, it has long been planned to make the author notificatons more centralised to provide more consistent reporting. Due to some very heated threads recently, David Golden and myself decided we should look at putting a centralised email alerting system into place sooner rather than later. David has been looking at alerting testers to broken environments, old CPAN Testers tools and badly formed reports, which will hopefully be live soon, while I have taken on the first part of the whole author notification and preferences system.
The prototype for alerting authors went live this lunchtime (UK time), though if you haven't received an email yet, don't fret it is just likely that none of your distributions have been tested since Saturday in order to receive the mails. As this is just the first part of the full preferences system, I've had to set some defaults. These have been set to notify authors of the FAIL reports, which are the FIRST instance of a distribution/platform/perl combination. Currently all the known PAUSE ids have been entered into the database with these defaults, apart from a few authors who have expressed other default preferences. If you do not wish to receive these alerts, or would like to alter the preferences, please send me an email (barbie[at]cpan.org) and I'll make the updates as requested.
The notification system will run once a day and collate a list of the links to the reports via the web interface on the NNTP server. As such, you will no longer receive huge reports in your inbox, but at most just one email a day containing a list of links to your reports. You can then view them at your leisure. If you have any problems with the notifications themselves, please let me know, and I would welcome any suggestions for improvements. I may not be able to implement them straight away (and may already have some on my TODO list), but please feel free to email me privately, comment here, or join the cpan-testers-discuss mailing list and join the dicussions.
[1] If you can think of a suitable word beginning with 'L' to replace 'Service' let me know ;)
One FAIL per platform per Perl version. Awesome.
Except that looking at my daily FAIL report, most of them are non-production releases...
ORLite-Mirror-0.06:
- i386-freebsd / 5.8.8 patch 34313:
- FAIL http://nntp.x.perl.org/group/perl.cpan.testers/2180772
- i386-freebsd / 5.8.8 patch 33662:
- FAIL http://nntp.x.perl.org/group/perl.cpan.testers/2180771
Can you please filter that crap out as well?
Adam K
Re:Needs a little more culling
srezic on 2008-09-12T18:07:37
Why crap? 5.8.9 is near, and we should find potential errors there as early as possible, shouldn't we? At least before the release is done?Re:Needs a little more culling
chromatic on 2008-09-12T19:30:26
The utility of those reports depends on who has responsibility for the failure.
Re:Needs a little more culling
srezic on 2008-09-12T19:56:56
So what if Nicholas decides: 34313 is good, nobody complained, so let's make 5.8.9 out of it? The report would at once become good, not anymore crap?Re:Needs a little more culling
chromatic on 2008-09-12T22:31:10
If 34314 breaks everybody's distro, what's the point of notifying every distro author? The problem's clearly with 31314, not half of the CPAN.
Re:Needs a little more culling
srezic on 2008-09-13T05:32:49
"Half of the CPAN"? Don't be ridiculous.Re:Needs a little more culling
Alias on 2008-09-13T06:15:52
Or in the case of "blead breaks Module::Build", 10% of CPAN.
Re:Needs a little more culling
srezic on 2008-09-13T06:31:43
May be, but this is not blead. This is maint.Re:Needs a little more culling
chromatic on 2008-09-13T07:41:43
Mistakes happen. That's why we have tests.
As a substitute for 'Service'... although I personally don't like the army reference too much.
Re:'Liaison' is good
mr_bean on 2008-09-13T11:05:21
Liaison also means an illicit sexual relationship, ie an affair.
http://www.thefreedictionary.com/liaison
Re:'Liaison' is good
polettix on 2008-09-16T22:28:17
In the true spirit of "make love, not war", I'm not particularly bothered with *that* reference
:)
