/ domm
Thanks to yi.org and the incredibly helpfull Tyler MacDonald I proudly present the new CPANTS site !
While things look mostly the same on the surface a lot has changed beneath, and even more will change. I'm now using DBIx::Class as a ORM (and still using Catalyst), Module::CPANTS::Analyse and the still unreleased Module::CPANTS::ProcessCPAN, which is build to allow incremental testing (i.e. test only the dists released since the last run).
It will still take a bit of work to get the incremental testing ready. E.g. I want to save some condensed stats of old results so that I can plot the kwalitee evolution of dists during time.
Another open issue (and I'll appreciate any ideas) is how to handle the ranking of authors in the cpants game. Currently it's based on average kwalitee. I would like to include the number of dists into the rank, because it's harder to get high kwalitee if you have lot of dists. Any ideas?
Original: http://use.perl.org/~domm/journal/29624
I find the site really useful for seeing if I can improve anything.
Just one question -- What's the "bad permissions" bit mean? I noticed it for my XML-Genx [perl.org] and I'd like to fix it...
Thanks,
-Dom
What's the "bad permissions" bit mean?
There once was a metric called 'no_bad_permissons' which checked if the dists only contains files writeable by the user. Because I hate it if after doing a manual install I end up with a dir I cannot remove (without doing 'sudo rm -r dir'). The metric was dumped as several people objected. The metadata remains...
Where to for bug reports....
Note the following entry in the Module::Install distribution "modules and pragmas used" section.
$self->{path}::$self->{dispatch} 1
I notice that the metrics quantity is the same... Any word on if or when we might expect some of the newer proposals on the QA list to be added?
Also, I think it will be interesting to see how you could deal with module quantity in the game, especially when you look at me/miyagawa/audrey and 100+ modules. You could easily end up throwing the numbers off dramatically.
If a fixed amount, it means you need to only give something like 0.01 or 0.02 points per distribution, which doesn't really seem like very much.
One minor issue, at the bottom of the pages you have text:
CPANTS data generated with Module::CPANTS::Analyse
but Module::CPANTS::Analyse is then a link to Module::CPANTS::Generator on search.cpan.org.
From memory previous versions gave details (somehow) on the exact causes of errors, for example in earlier versions of Imager I could see it list lib/Imager/Cookbook.pm as not having use strict. Unfortunately I don't remember exactly where that was linked from - is the detailed scan information available
I have a few modules that use Test::Inline 2 that also get caught up in this.
In my case, it's due to not regonizing the (not yet common) "extended begin" syntax.
=begin testing foo after bar
=end testing
... on Atom!