All the Perl that's Practical to Extract and Report

Filtering of CPAN submissions is a terrible idea.

petdance on 2005-07-13T22:16:52

London.pm is currently thrashing around the whole idea of "We shouldn't allow modules on CPAN unless X" idea. X can be anything from "passes some arbitrary tests" to "gets OKed by some group" to anything in between.

CPAN thrives BECAUSE we allow unfettered uploading of shit, not in spite of it.

We should work to make what is there better, and to help people find what is good, rather than making arbitrary exclusions.

All this started because of Hello-0.01 getting uploaded. Sure it's useless, but so what? Is there any tangible harm? "It's using a top-level namespace!" is not harm, and if it was, folks on modules@perl.org could take care of it. It's only annoying if you allow it to annoy you.

All the ideas of "Let's filter based on X" are just solutions in search of a problem.

Agreed

davebaker on 2005-07-14T01:45:21

Perl already suffers from a perception of being "hard-core"; PHP has an image of being "easy-peasy" and attracts many young programmers. Let's give the new programmers an opportunity to see their name in lights (on CPAN), even if they (like most programmers) eventually shudder about over some of their code from earlier days.

But regarding the bigger issue of quality control on CPAN, how can we reward/highlight/promote the best of the bunch?

The Phalanx project is a big step forward in that direction, I think, because it helps me know which modules are the "biggies."

Let's do even more.

Isn't that....

speters on 2005-07-14T02:06:14

what CPAN Ratings is for. If you don't like a particular module, voice you opinion. If the module is bad, chatting about on IRC without providing feedback to the author is next to worthless.

Agreed

jmcada on 2005-07-14T05:02:10

Don't filter CPAN. Only knock modules off if they have malicious intent.

Instead, create bundles of the best modules. Phalanx has a bundle... anybody can post a bundle... don't filter out crap, point out the shining stars.

A step further than that, I see a real business opportunity in creating a filtered mirror of CPAN modules. This mirror can contain modules that are useful, safe, and proven via tests and so fourth. Corporations will buy into a 'trusted' and 'supported' set of modules on CPAN. This has the potential not only to earn someone some scrilla for the implementors, but could even give Perl some corporate/commercial support. Like it or not, that is where the money is.

Re:Agreed - Highlight the best

TeeJay on 2005-07-14T08:37:16

I'd be quite interested in a commercially supported core set on modules. I'm sure that could be very useful when you have smaller businesses with only 1 or 2 programmers and no time to evaluate, debug and test CPAN modules.

The ratings site is quite useful, if underused. I'd like to see some kind of organised best-of-breed site that combines recomendations, with pointers to articles and feature comparisons...

Yes I was on #london.pm, yes I too muttered about bloody stupid modules, but also the number of times many of us have uploaded a module 2 or 3 times in a few minutes because everybody makes mistakes (and usually spots them a second after the module arrives in the PAUSE queue and is announced everywhere).

Feeding off the idea mentioned on irc about people responsable for namespaces, it would be nice if you had somebody who put together a comparison/guide to particular areas handled by CPAN modules, with best of breed, articles, books, mailing lists, etc. That would really "add value" to CPAN.

Re:Agreed - Highlight the best

GAVollink on 2005-07-26T17:31:37

Late in mentioning... but doesn't ActiveState fit the "commercially supported core set on modules". They certainly sell themselves that way (even if I only use them on WinTel).

On the topic of stupid modules... User naiveity, more often than not, is the cause of crap. The whole "My situation is unique, so I'll code and upload my own quick and incomplete solution without looking to see if something else comes close," attitude.

Luckily, by virtue of the amount of energy it takes, it's relatively rare. At least more-so than good idea, unique modules that get uploaded and become stale due to lack of time.

ironic

hfb on 2005-07-14T06:45:55

when one considers that much of the Acme namespace comes from London not to mention it's namesake is a memeber of London.pm as well. The depths of uncleverness found in the Acme namespace alone far outweigh some poor sod who wants to upload his first module unwittingly though I blame certain perl books that encourage the readers to upload their homework to CPAN.

OMG CRAP

sungo on 2005-07-14T13:54:40

This may surprise you but I totally agree. CPAN is full of crap. Not just intentionally useless crap, there are lots of broken, unmaintained modules out there. What oh what should we do about it? There are two reasonable actions here. 1) In the case of amusing but useless modules (including those you don't find amusing but someone else does), fucking get over it. 2) In the case of unmaintained, broken modules, take it over. We all should be getting off our asses and trying to make the CPAN a better place, code wise, if it really bugs us this much (preaching to myself here too).

One of the commenters suggested Bundles of useful modules. Someone else pointed out cpanratings. I'm pondering a blog, use.perl feature, perl.com regular article, or something else along those lines that could just exist to point out useful and well maintained modules. Social engineering of the right sort can go a long way here.

Don't change CPAN, just change your view of it.

schwern on 2005-07-17T00:57:07

Build your own index of modules (CPAN/modules/02packages.details.txt.gz) using whatever criteria you like and make it publicly available. Hack the CPAN shell to get its index from your site while downloading from a real CPAN mirror.