Lets be honest, there is a lot of crap modules being developed. Not just crap, but superfluous ones, and ones that have had two commits by the author who has then gone AWOL. The contrib pile is a mess right now.
As Drupal rightly gains momentum, more people will submit modules, often that do the same thing as another in a worse way.
We need some kind of filtering system, this site is great, and it goes a long way toward fixing what the Drupal.org project listings should have done in the first place.
But it relies on user votes to reach its full potential. What would be useful is if the site built its own rating for each module. Create an algorithm that looks at module 'activity'. By that I mean, commits, downloads, module size and probably more importantly activity in the issue queue.
Checking for closed issues, author replies, patch submissions etc. From that it should be fairly obvious which modules are the most popular and in theory be of a higher quality.
Don't know if its even possible, but something needs to be done!
Comments
Metrics
John - August 13, 2009 - 09:02
Activity-based metrics are a popular topic of discussion on groups.drupal.org. In fact, Drupal.org recently started displaying issue queue and commit stats right on the project pages.
However, I don't believe these raw numbers can tell you much about actual code quality. Here are the problems I see:
The number of open issues largely depends on how popular your module is, and what skill-level your module targets.
Number of commits is distorted by developers who constantly commit small changes vs those who just write something that works the first time.
Even module popularity can't tell you that much. Some modules are just guaranteed to be popular based on their name. People will download and install poor quality modules if there are no visible alternatives.
My hypothesis is that anyone trying to score modules based on this data will find there's a lot of bad modules with stats that look just like good modules. It might be possible to find a few good predictors of code quality, but in the majority of cases, it will be tough to distinguish who the winners are.
Your right of course, its
g3f - August 18, 2009 - 13:37
Your right of course, its always going to be flawed, a new module might be the pinnacle of programming skill and still be at the bottom of the pile with one commit and zero downloads.
But its not meant to be a defacto way of finding out which module is 'best'. Its just another way of filtering results. Much like you already have most downloaded (being able to sort search results by most downloaded would be ace by the way) you could add extra, optional, ways of filtering.
Sort order
John - August 18, 2009 - 20:43
"being able to sort search results by most downloaded would be ace by the way"
Module Finder search results are already sorted by downloads. It used to show results in alphabetical order, but I changed it a few months ago because sorting by downloads produces better results.