Tracy R Reed wrote:
Andrew Lentvorski wrote:
Functional progrmming doesn't need a cheerleader. It needs a problem
to which it is an actual solution.
It seems Erlang was the right choice for Ericsson's phone switches. And
for jabberd. I think Lisp was the right choice for emacs. And as the
number of cores grow in our cpu's I think we will find many more
problems to which functional programming is an actual solution. Can you
do hot swapping of code in any non-functional language? The paralellism
just seems to fall out nicely when using a side-effect free language as
well.
Hot swapped code? Sure, it's trivial. Here's a snippet from FreePascal:
MyObject.SomeProcedure := @SomeOtherProcedure;
This replaces the procedure that was originally associated with the
object MyObject with the replacement procedure SomeOtherProcedure. It
does this by just swapping out pointers in the virtual method table. The
only restriction is that it has to have the same signature (type and
order of parameters). Because of this, you can store a bunch of routines
in a shared object (*.so) and use dlload to provide the binding. Your
program still calls MyObject.SomeProcedure but you can get totally
different code execution depending on what you bind to the procedure call.
You can even store entire classes in a shared object and dynamically
load/unload them, so it's feasible to have a same-named class do
different things depending on where you got it from. This could be very
useful where you want "business rules"[1] customized to each customer
but want all the calling conventions to remain the same.
Gus
[1] I use this term only in the sense that there are no polite words to
describe the steaming pile of dung that some people think is logic.
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-lpsg
