metrics for cpan

[Kirk Kimmel]

We already have all kinds of interesting data about CPAN and what it does
but I could not find something. I have been submitting cpan reports for
over a year and I just started wondering how long does it take to test all
modules on CPAN once, not including versions of Perl?

I have been trying to keep an eye on my newest vm just to see how long
things take. Right now I estimate a normal hour gets 100 reports done but
some hours are much heavier.

I think having some public facing data about testing times could be useful
to the community and this is the first thing I thought of.

Kirk

[David Golden]

I haven't smoked all of CPAN lately, but I believe that Steffen
Mueller smoked it against bleadperl recently and it took several days
but less than a week.

The thing about typical test reporting rates is that it depends on the
rate of new uploads, the size of dependency chains (if those have to
be built/installed), and the runtime of the test suites.

That's different than smoking all of CPAN, where many of the older
distribution have smaller dependency trees and smaller test suites (if
any tests at all).

They're related, certainly, but I suspect smoking all of CPAN will
give you a higher reporting rate than you'd experience just smoking
recent uploads.

David

--
David Golden <x...@xdg.me>
Take back your inbox! → http://www.bunchmail.com/
Twitter/IRC: @xdg

[Chris 'BinGOs' Williams]

As I use minismokebox I have smokers set up to use the IRC plugin so
I can monitor their progress. (App::SmokeBox::Mini::IRC)
<http://cpanidx.org/heh/smokeboxinit.jpg>

I just did a little trawl back through the scrollback buffer and generated
a CSV <http://cpanidx.org/heh/smoker_data.csv> of the data gathered.

Some explanations:

The hosts beginning with 'd' are new DragonflyBSD hosts that haven't
smoked before, so will have no previous test reports generated so will
be running test suites for everything.

The others are long running smokers, so will skip tests for previously
passing test suites.

totaljobs, idlekills and excesskills are measured in 'jobs'. A job is
an individual distribution uploaded to CPAN.

idlekills are when the job is killed due to 15 minutes of interactivity.
excesskills are when a job runs for too long ( over an hour I believe).

average, minimum and maximum are durations in seconds.

I did write another plugin for minismokebox, called funnily enough
Stats (App::SmokeBox::Mini::Plugin::Stats), but it was generating huge
SQLite databases, mainly because I had forgotten about it, and I lost
interest in doing anything with the data, so I stopped logging that stuff.

--
Chris Williams
aka BinGOs
PGP ID 0x4658671F
http://www.gumbynet.org.uk
==========================

[David Cantrell]

On Thu, Nov 29, 2012 at 08:56:05PM -0500, Kirk Kimmel wrote:

> We already have all kinds of interesting data about CPAN and what it does
> but I could not find something. I have been submitting cpan reports for
> over a year and I just started wondering how long does it take to test all
> modules on CPAN once, not including versions of Perl?

It Depends.

How much hardware will you throw at the problem? Do you test, install and
keep dependencies? How long do you keep them for?

--
David Cantrell | London Perl Mongers Deputy Chief Heretic

Languages for which ISO-Latin-$n is not necessary, #1 in a series:

Latin