I think it may turn out that the real genius of Scala (and Clojure) may be in discovering the only truly viable model of real-world adoption for a functional language - piggybacking on a bigger, dumber, mainstream success. Just consider what Scala gets for free that Haskell has to build itself: a fast, robust VM, libraries for any mundane thing under the sun, and a foot in the door at any common dev shop.<p>The real cathedral vs bazaar issue here is deeper than the library issue. It's the runtime that counts and sharing a runtime with something like Java delegates all the grunt work to the legions of Java programmers and lets Scala hitch along for the ride picking up all the interesting work on the top of the heap. The future of functional languages is in symbiosis.
This article got me thinking a little. I know base pretty well; whenever I'm writing some code, I always end up hitting the "source code" link and checking out the implementation. I never really wonder "how does this work" as I'm programming anymore.<p>The question is, how can one go from knowing a library to contributing to the library? One of the problems with HackageDB is that you can see everything, but you can't see where you can make changes.<p>I had a problem with C2HS a few months or so ago; it didn't work on Windows (because DLLs use "stdcall", not "ccall", and the "ccall" string was hard-coded). I clicked the darcs link on Hackage to get the source. Dead. So I just edited the source on my local machine and moved on.<p>A few months later I noticed that darcs was working again and there was a new maintainer, and I did get to contribute my patch. But it was kind of dumb luck, I could have easily had never noticed, and this bug never would have been fixed.<p>So right now, it's easy to find libraries, it's easy to understand libraries, but it's not always easy to improve the libraries. And incidentally, the biggest problem with libraries in Haskell is that they never get out of the "rough draft" stage, by getting contributions from people other than the author.<p>Incidentally, I don't have this problem with Perl, because everyone happens to use Github. This has created an interesting effect where libraries that aren't on Github become very difficult to work on, and people seem to favor rewriting instead of fixing things "the old way". (diff + an email to the author.)
What sucks is that Hackage has been down all day (and is down not infrequently) and I've needed it desperately. Also, Don Stewart's (evangelized by OP) 'Download' package won't build on OS X, with a different cryptic error message than it gave me last week. Yes, these are piddly little whines of personal anecdote, and objectively it's absurd to suggest should be flawlessly and continuously taken care of by the Haskell establishment at this stage in its development, but Haskell is pretty quickly becoming my language now:<p>I love it, I think in it, it is a daily struggle to fight the urge to evangelize like a douchebag. With the problems I've just mentioned, I'm writing what I need to in Python, hideously functional Python that would make Guido cry, but prototyping everything in Haskell. If this stuff is pissing me off, a momentarily curious Ruby bro sticking his neck out and cautiously circling the >>= in the stone is going to get bit, fuck off and not look back.<p>Please build the cathedral soon.
Wow, I started reading this without looking at the bottom to see what the proposed solution was. It turns out there is none.<p>Still, it was pretty interesting for the quick analysis of the various ways that an open source programming language's basic libraries are currently developed.