One tedious but helpful approach is to produce XS interface modules to new C libraries. It's pretty easy to track down through CPAN whether one has been done and do it if it hasn't.

XS is maligned these days, but it's how P5 works. It is incredibly useful to understand.

Hmmm? I really don't think that would get that much use. Most of the time people ask, "hey why isn't there a function for...." and they will usually get an answer for a certain module in cpan.

If you have something in mind, why not ask? Chances are somebody has done it. If they haven't, set up a project. If you want help, I am sure others would help.

So I suggest that there be a section listing modules that should exist, but do not yet.

What's with this "I suggest that there be a section for..." frenzy, as of lately? I'll tell you in advance that such requests tend not to meet popular approvation, and someone will certainly give you more detailed info on why it is so. Personally I'll only add that IMHO far more interesting proposals for new sections have been given - and they were rejected nevertheless: what you need is addressed perfectly fine by Meditations. OTOH this convinces me more and more that what we need is a tagging system.

There's lots to be done in existing modules. WWW::Myspace is one that not only could use more features, but needs to be constantly updated to match changes at http://www.myspace.com.