[comp.lang.forth] Min and max

wmb@MITCH.ENG.SUN.COM (Mitch Bradley) (06/05/90)

> An even better example [illustrating minimalism vs kitchen-sinkism] is
> memory allocation packages.  ... there is no best memory allocation scheme.
> ... The Forth literature has examples of all types of storage management
> mechansisms that can be added to Forth systems.

Except that "adding a storage management mechanism" in a portable fashion
requires the ability to acquire the storage that is to be managed.  Under
most operating systems, an individual task does not "own" all the memory
of the machine, so Forth can't just use any memory that it wants without
asking first.  Furthermore, it is not in general possible to use ALLOT
to request memory from the OS, because ALLOT manages a guaranteed-contiguous
space, and memory acquired "on-the-fly" from an OS cannot be guaranteed
to be contiguous with the memory that is already managed by ALLOT .

The memory allocation wordset provides a minimum set of capabilities
necessary to dynamically allocate OS memory.  Other memory allocation
mechanisms may be built on top of the provided wordset, but without the
basic wordset, those additional memory allocation mechanisms would be
"banks without any money" (i.e. savings and loans :-).

> minimalists would like to preserve the fundamental language building tools
> ... Mitch Bradley would like the glossaries of certain popular add-ons
> specified.

There should be the word *also* after the word "Bradley" in the above
sentence.

I have done my share (probably more than my share) of committee work in
an attempt to preserve the fundamental language building tools.  In
particular, my proposals for PARSE , COMPILE-TOKEN , GET-SEARCH-ORDER ,
and the proposals precisely detailing the semantics of file loading fall
into that category.  Time after time I have argued for exposing the
primitive functions necessary to implement a feature.

So, I claim that in my role as a "kitchen-sinkist", I am NOT GUILTY of
the "mistake of trying to standardize the language at too high a level".

> There really is no conflict [between minimalism and standard extensions]

I agree.


Mitch Bradley