Good question!
I think it would be possible to do both. You could also
group by author and not count references when you include your own module/distribution in your own code.
The problems with the using CPANTS use-by for my purpose is 1) you need to know what distribution a module you're interested in lives in and 2) you need to know which of the used-by results actually use the module you're interested in.
Remember, I'm not looking for a way to rank authors or the value of a module. I want to help folks deciding what module to use from among a list.
For instance, if you've never written a script that interacts with an IMAP server before and you need to do so now, do you use Net::IMAP or Net::IMAP::Simple? I want to make that decision easier.
| [reply] |
if you've never written a script that interacts with an IMAP server before and you need to do so now, do you use Net::IMAP or Net::IMAP::Simple?
But then basing your decision which module to use for a script depending on the usage in other modules is quite fishy.
Take for instance Getopt::Long. IMO, a wonderful module, and probably used in a gazillion of places. Less useful in modules. It gets quite a number of "uses" by other dists on cpants.perl.org, but judging on the names of the distros, many of them provide applications.
Furthermore, there was a time that if you'd let number of other people decide which module to use, you'd use the stuff from the infamous Matt's script archive.
Popularity is its own machine. If everyone uses the same principle, once a module is on the top spot, it will always remain there - because since it's on the top spot, it will be used more often than other modules.
| [reply] |
I don't think it's fishy, but that's my opinion. If lots of people use it in modules, it's probably not crap.
The idea is not to dictate what modules people should use, but rather to help narrow the field when there are lots of possible choices to choose from.
Plus it might be fun...
| [reply] |