Social CPAN : Finding the best and most popular modules

[jannino]

CPAN should be a popularity contest. Why? There is tons of great stuff on CPAN, but knowing what is considered the best, most used, most stable, most loved, can be hard. Lots of people use search.cpan.org, but there needs to be better easier ways to capture their knowledge for the benefit of others.

There is a rating and reviews system, but most stuff has no rating or reviews, and you can't sort the results by ratings (as far as I can tell). Also the ratings interface isn't available from the pod doc pages of modules, only from the main page for the distribution, which is looked at less often. The review information for the distribution should be accessible in the right hand bar of all modules in the distribution.

But to get good data on usage and quality quickly, there should be more than that. What if you could create a user account on a cpan frontend website. You could use it to:

• Star distributions as favorites
• Get a homepage listing your starred modules and any news relating to them
• Quickly give a star rating and/or review to a distribution
• Get popularity ratings on distribution based on other user's stars and ranks
• See popularity based on the number of other distributions that depend on a particular distribution
• Include any of the above in searches
• Use a command line tool to submit information about what modules you're using in a given project
• Usual social network stuff, like linking to friends and professional contacts.
• Relate your account to accounts on other perl sites, like this one, so people can find you in all the different corners of the perl universe.

The distro level tools should always be available on module pages, in some mini form in the right hand nav, since those are the pages people are really looking at when they come to CPAN.

There has been a lot of talk about getting some "buzz" going about Perl, and this sort of thing just might help. CPAN is a great resource, but it can be a bit hard to deal with its breadth. I think a lot of great things have been done around annotating things on CPAN so far, and something like the above would just make it better, and help bring it all together, and maybe get some of those tools used a little more.

This is my first post here. Hello! :)

[SankoR]

What you propose seems a little more all encompassing than most ideas to modernize the CPAN (and related services) experience. But hey, since you're bouncing ideas around, if you haven't already, you may want to drop by the rethinking-cpan mailinglist Andy created (Update: Check out his related write up, too.). You may receive more focused feedback there and, once you're ready to put your idea into practice, that would probably be the best place to recruit help.

And for more on the pain of the CPAN Ratings system and a mini pro/con debate on "improving" CPAN in general, pick through this conversation from the perl.qa mailinglist.

[jannino]

Thanks for the pointers. I will check all of them out.

Thanks everyone for the replies they are great.

For ideas on where I am coming from:

• Ratings: Netflix does this really well and simply, as do many online retailers
• Starring:: Google reader lets you star, share, tag, and email articles. I find it very useful to star articles so I can find them quickly later. Being able to do that to my favorites modules would save me lots of time getting to the docs I need. I have to use del.icio.us for that right now.

I wrote the post to be a bit provocative, since I wanted responses. Of course choice of modules for a project is a complex decision that takes some research and asking around. CPAN I think should work as a way of shining a bright light on what there is some consensus on as the best tool for the job, if you want it to, but it should never hide anything. Any new features should be of course take it or leave it, and should be easy to use, and just as easy to ignore if you don't find them useful.

[merlyn]

There is a rating and reviews system, but most stuff has no rating or reviews
...
What if you could create a user account on a cpan frontend website

What makes you think your system would be used any more than the existing ratings/review system, which today requires only a minimal registration?

I fear the problem here isn't mechanism... it's motivation.

[tye]

On several occasions I've been motivated to post a review or to vote on a review. In every case, my motivation was unable to survive the roadblocks of the "mechanism". I don't recall what the specific road blocks were. I just recall more than one way that the implementation showed me an interface to invite me to do something and then changed its mind and denied me. Having tried more than once to jump through the hoops to satisfy its login (and perhaps other) requirements, I have yet to succeed.

I don't even recall what the requirements were, but had they actually been "minimal", I'm sure I would have met them (having met registration requirements before). I won't dispute that the requirements are likely "small", but they have certainly been enough to stop me despite multiple attempts.

Perhaps I'm just an abnormally lazy or picky person, or something. Or perhaps the implementation has some room for improvement. I find requiring registration to add a review for a module to be a rather disappointing statement on the implementors' view of their fellow humans. *shrug*

[GrandFather]

We are Perl programmers - we are supposed to be lazy.

I've had exactly the same experience and remember just as much about the roadblocks. *sigh*

[JavaFan]

I don't care about ratings. For me, there are only two classes of modules on CPAN: "modules that solve the problem I needs to fix in a reasonable way" and those that don't. A module M1 that is stable, lightning fast, beautifully formatted, documented in perfect prose, has and has tons of features is less useful to me than module M2 that has no documentation, a clunky interface and horrible code if M2 solves a problem I have, and M1 solves a problem I don't need to be solved.

[jeffa]

"CPAN should be a popularity contest."

No it should not be. In fact, most programmers choose this line of work because they want to get away from such MARKETING ploys. Personally, i think CPAN is just fine the way it is. It is not broken. If you have a hard time finding modules then maybe you need to address that problem within yourself, because i have found that finding modules is not hard. Asking questions at forums like this is not hard. Knowing where to go to find the answers is the margin between a great programmer and one who merely found themselves in this line of work. Increase your education. Don't turn everything into a competition.

[Lawliet]

I was going to comment on that quote too. The example I was going to use, however, was 'Look at PHP'. I learned some time ago that popularity does not equal greatness, usefulness, or anything, really.

[mr_mischief]

I don't care which CPAN modules are the most used. I care which ones help me get my work done quickly without adding a bunch of bugs. A feature matrix of similarly targeted modules makes much more sense to me than ratings.

[zby]

These are nice ideas. If you'd like to try them out and write some code then we infite you to he CPANHQ project which attempts to be an Open source rewrite of the CPAN module search/browser. It's still in early stages - but there is enough code to import some modules data and try out new ideas. The problem now is of bootstraping the system, we cannot assume we have all those additional socially gathered pieces of module data - we need to start with the currently available information, make something useful out of it and only after people find it compelling enough to switch from the two main cpan search engines - only then we can count on them adding new data.

[Khen1950fx]

I've been working a CPAN-auxilliary tool script. Why don't you try it. You can get it here: cpanfm. It still has a long way to go, and I haven't started writing the module yet; also, some of the commands are misbehaving. If you like it, please let me know. It's lightweight, very fast, and pretty easy to use.