On Sep 5, 2010, at 7:46 PM, Mathew de Detrich wrote:
Another thing you can say is that Perl is a very extreme language in design where as Haskell is more "general". This means the one thing Perl does, it does very well (expressing programming problems in the most concise/short possible way) but it has to sacrifice for it massively in other areas which end up costing much more in the long run. Most 'real' world problems do not require that amount of brevity, considering the massive cost that Perl brings for such a thing.
That doesn't sound right to me. Perl's biggest weaknesses are traditionally: (i) the syntax: but those $'s and @'s are actually type annotations; and (ii) "There's More Then One Way to Do It": the existence of multiple approaches to solving a problem, instead of an "official" obvious choice. This means that every programmer on the team either has to KNOW all the possible ways to solve a problem with Perl, or the programming team has to CHOOSE one and make it "policy" -- effectively picking out the nicest bits and sticking to that sub- language.
Depending on your point of view, Haskell does not compare particularly favorably with respect to "TMTOWTDI". The whole Control.* hierarchy is the construction of custom control structures. That's the whole point of "glue" languages. You write custom control structures to support the chosen normal forms for expressing data and computations.
_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
