Top 100 site adds the "Debian Most Wanted" list

Alias on 2009-02-15T15:58:16

http://ali.as/top100

The "Debian Most Wanted" list is my first attempt to mix in kwalitee metrics into the lists.

In this case, it takes the kwalitee metric distributed_by_debian, inverts it into a decimal 1 or 0, then multiplies it by the modules volatility score, in effect just filtering the list, but doing it more complex under the covers to experiment for later more complex indexes.

The numbers currently look to be extremely flawed, at the very least they ignore dual-life modules (I'll need to see if there's a way around this in the SQL, since all the lists are created purely from the database for reasons that will become obvious later).

There's also a ton of modules listed that I know for a fact DO actually have Debian modules, for some value of "Have Debian modules".

To deal with some backlash I'm starting to get about the lists, I've also put a large prominent disclaimer on the front page to indicate I'm not responsible.

Hopefully this will act as a recruitment channel for people to help domm out with the kwalitee metrics data :)


Even they don't know

gabor on 2009-02-15T16:23:21

One of the problems with the data is that the Debian people cannot supply the full list of the perl modules.

Maybe this list will get them work a bit more on their end so we will have a better list.

Re:Even they don't know

Alias on 2009-02-16T01:14:45

I wasn't aware that the name was manually supplied by the debian folks.

If this turns into a terminal problem I'll probably kill this list.

Re:Even they don't know

gabor on 2009-02-16T04:47:47

AFAIK some of the packages are maintained by the Debian Perl group - those are in the list - other are still maintained (or not) by individuals. These have other repositories and other formats and thus most of them are not in the automatically generated list.

They are in the long and slow process of converting more to group maintenance.

Filtering out core modules?

brother on 2009-02-17T07:22:59

Nice list, if we can manage to filter out the false hits. Filtering out modules shipped with perl would be a first step.

I hove no idea if these modules should really have the 'distributed_by_debiabn'-kwalitee set.