On Tue, Nov 13, 2007 at 11:57:30PM +0100, Benjamin Teuber wrote:
> On Nov 13, 2007 10:18 PM, Nick Apperson <[EMAIL PROTECTED]> wrote:
> > With the next generation of C++ with
> > variadic templates I think C++ may overtake Lisp for metaprogramming, but I
> > don't know enough to really make that claim.
>
> I don't know about variadic templates, but in general it is almost
> impossible to overtake Lisp in terms of metaprogramming, because every
> time a new cool programming technique is invented, some guy will
> implement this in 10-50 lines of Lisp code and voila - Lisp now has
> it, too =)
There is a considerable amount of truth in that. Usually someone
has to see it to believe how CL is flexible enough to support things
like hand-rolling things like your own object system, only Forth in my
experience brings the same "you can do *that* too?!" level of surprise.
Still, though, sometimes useful things resist elegant implementation
in CL. For example:
* I doubt CL can be customized to support a thread library nearly as
gracefully as languages designed around threading from the
ground up.
* For complicated code where a very expressive static type
system for higher order functions is a good fit (like the SML
code in Umut Acar's thesis) CL is still typesafe, but the way
CL postpones much of the error detection to runtime can be a
significant source of friction (and loss of expressiveness, as
code doesn't say as clearly what it is doing). (I'd much rather
translate that SML library into CL than into C++, though, though
I admit I know very little about Boost except that it's designed
to help with that sort of thing.)
* I don't know any way of expressing functors in CL that's as
nice as languages which have them built in.
Also, to the considerable extent that that is true, note that it
carries a sting in its tail. CL is so insanely flexible because
fundamentally a CL program is not a declarative blueprint of the
finished program, but an imperative recipe for building the finished
program: a CL program is not as much like a C program as it is like a
software package which can include C programs which write other C
programs (e.g., software packages like lex or yacc or the old cfront).
So you can express anything, but your program analysis tools
(debuggers, profilers, code coverage tools, profilers, automatic
refactoring tools...) may be hard-pressed to keep up in an
understandable and/or reliable way, for pretty fundamental reasons.
--
William Harold Newman <[EMAIL PROTECTED]>
PGP key fingerprint 85 CE 1C BA 79 8D 51 8C B9 25 FB EE E0 C3 E5 7C
Ubi saeva indignatio ulterius cor lacerare nequit. -- Jonathan Swift's epitaph
_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
