I finally got around to installing perlbrew, 4 flavors of perl and updating some modules for recent perls. I've been sending a lot of email test reports lately.
Last night I got the email from bitbucket about CPAN Testers 2.0 and the death of sending reports via email with wiki links/instructions on how to setup .cpanreports/config.ini for a metabase id/transport via HTTP.
This is all well and good, but it feels wrong to me. I use CPAN::Mini. The point of which is partly to have CPAN when you're not online. The same reason I use git. With SMTP emails I could just queue up test reports in the local postifx and they would get delivered later when I'm online.
Now, that's is no more. I'm forced to be online to send HTTP reports. Seems like a bad idea. NOw, the only thing I can do is toggle off reporting when I'm offline, which really means I'll hate doing that and just turn off reporting all together.
Why did reports via email have to go away?
... we can add offline support for http without TOO much difficulty.
Re:Caching doesn't need email
link on 2010-08-10T08:11:05
I guess that is progress. We have gone from supporting offline reports now to supporting offline reports in the future and all we have to do is wait from someone to write code.
Re:Caching doesn't need email
Aristotle on 2010-08-11T04:57:49
Yes, that is progress. The servers running the mail load for CPAN Testers 1.0 could not keep up, and the machines are graciously donated and the system administration is a volunteer effort. They could not afford to continue providing services to Testers 1.0; the mail pipe had to be turned off. This is why Testers 2.0 was designed – not because some people got an itch to engage in architecture astronautics or arbitrarily decided that email was passé. Whatever you think of the new system, the old one is going to stop working, no ifs or buts.
If you ask me, a system that will continue operating clearly beats one that is about to stop.
Re:Caching doesn't need email
Aristotle on 2010-08-11T05:01:02
Oh, and the people who built Testers 2.0 put in a lot of overtime to ship it in time before the Testers 1.0 kill date.
Re:Caching doesn't need email
jk2addict on 2010-08-10T12:48:33
Sure, caching offline doesn't NEED email. But emails servers have been doing it since the dawn of time. So now we have to reinvent the wheel.
Re:Caching doesn't need email
chorny on 2010-08-11T08:40:37
E-mail was too complex here - processing of e-mails requires much more resources than processing of CPAN reports. So requirement to drop e-mail was not invented by cpantesters - it came from cpantesters "hosting" - perl.org.
... but indeed there has to be a way to collect logs offline.
That's what we use in OpenBSD to gather reports automatically for the modules in the ports tree.
So basically the same argument, except from another perspective.
http://search.cpan.org/user/bingos/metabase-relayd-0.18/bin/metabase-relayd
Email went away because the expense of operating the required services became to high for the people who were donating the machine-time for free, and because the technical debt inherent in the design was too high.
The relayd above is already in use by heavy testers.
Re:a relay exists
jk2addict on 2010-08-10T17:49:36
<sarcasm> Hurray! More complexity! <sarcasm>
Thanks for the link!:-)