How far can stack [LIFO] solve do automatic garbage collection and prevent memory leak ?

Alex McDonald blog at rivadpm.com
Wed Aug 18 20:49:11 CEST 2010


On 18 Aug, 11:09, spinoza1111 <spinoza1... at yahoo.com> wrote:
> On Aug 18, 1:21 am, Standish P <stnd... at gmail.com> wrote:

>
> > This you might want to take this to the Forth people because they are
> > marketing their language as a cure for all that plagues programming
> > today.
>
> No, they're not.

That I agree with.

> Stack based languages have seen better days and Forth
> (and the SL/1 language I supported with compilers at Bell-Northern
> Research) were last in fashion in the 1970s. Processors seldom could
> multitask, so it wasn't recognized that the stack could be a
> performance bottleneck, where stack operations cannot be pipelined or
> executed in parallel.
>
> John Hennessy of Stanford and MIPS made the stack must die case at ACM
> ASPLOS in 1987. Niklaus Wirth was also at this conference at which I
> was a fly on the wall, maintaining that the stack was good for
> reliability and verifiability of software.
>
> Forth had a snowball's chance because it forces ordinary programmers
> to think in Reverse Polish notation and is for the above reasons hard
> to pipeline, although of course it can be pipelined.

I really don't understand much of what you're saying here; Forth can
be implemented on processors that have several hardware assisted
stacks, 1 stack or even no stack at all. Multitasking? Why's that a
problem? And why is it hard to pipeline? Are you thinking of a
specific processor?








More information about the Python-list mailing list