Re: APL & Programming Concepts

Mb <mebian@[EMAIL PROTECTED]
> writes:
> Gosi wrote:
>> Something is not getting through to the general public.
>> In the very old days something could be objected against APL but I
>> have not seen any valid arguments against APL for many years and still
>> APL is not growing as much as it should.
>>
>
> Could this be that APL needs a standard, rock solid interpreter,
> supported and  in the public domain, with no bells and whistles but
> open to the outside world (C/C++, Fortran, libraries etc.), so who
> wants to start exploring can do it, and maybe a translator to C or a
> compiler, if the performances are to be boosted up for some
> application ? (My dream wishlist, I know :-) ).

I don't think the "translator to C" is likely to have anywhere *near*
the value that you might imagine...

The core of well-made APL apps tends to be in the form of pretty large
vector operations, and that's something that is already really heavily
optimized, and which translation to a "serial" language like C isn't
actually going to improve.

Not to say there can be *no* improvement; there has been work on such,
but just like recoding apps that do great gobs of string work from
Perl into C doesn't necessarily help much, since the string operations
and libraries were already heavily optimized, the same tends to be
true for vector operations...
-- 
output = ("cbbrowne" "@[EMAIL PROTECTED]
" "linuxfinances.info")
http://linuxdatabases.info/info/linuxdistributions.html
Those who do not learn from history, loop.