CPANTS Metrics

davorg on 2008-06-11T10:20:41

Gabor asked for feedback on his proposed CPANTS metrics. Here is mine.

I don't like them.

Well, I don't like most of them. I'm ambivalent on the Test::NoWarnings one.

It's the Debian ones I really object to. I don't like the Fedora ones that were added recently for the same reason. I don't think they measure Kwalitee (or quality) in any meaningful way.

Don't get me wrong. I think it's important to up package CPAN modules for commonly used packaging tools like rpm. I've done some work in this area myself. I'd really like a system where I can get statistics on which of my modules are being built for popular systems like Fedora and Debian. And getting information about why a particular module isn't being packed (license issues perhaps, or bugs) would also be very useful.

But I don't think this stuff belongs in CPANTS. I don't think it says anything useful about the quality of the module. I don't think that CPAN authors should be forced to care about platforms that they might have no interest in. It would be nice if they did, but I don't think you can penalise them if they don't.

So, for what it's worth, I vote "no" on the new CPANTS metrics. And I'd further like to see the removal of the other Linux distribution based metrics. But I'd really like to see that information made available elsewhere.


downstream info is useful

TeeJay on 2008-06-11T14:58:41

Downstream info is useful, but yes it's not directly related to Software QA in any sensible way.

The info is very cool, but needs to be somewhere else, essentially it looks like a new project.

The bit that is nice though...

Alias on 2008-06-12T01:07:50

... is the "easily repackable by" metrics, which I can see adding to software quality, because collectively they serve as a good scoring of "plays well with others".

I agree that "has been packaged by debian" is iffy.