On Mar 1, 1:52 pm, "Ed" <nos...@[EMAIL PROTECTED]
> wrote:
> If the header is compiled at the end of the definition then one will
> need another field - a cfa pointer - since the name length can no
> longer be used to calculate the xt.
> Apart from consuming an extra field, I'm not sure where the
> advantage is. Perhaps someone can explain it.
Headers building down from the top and code building up from the
bottom makes it easier to strip out headers. Indeed, if the headers
from a wordlist are saved externally in a group in permanent storage,
like a serial Flash RAM, you can only have the words present in a
system at any one time that you need visible.
> That view of forth is itself a sandbox.
And then proceed to put Forth in a sandbox, as to what it can't do,
based on a notion of what people haven't done in it:
> Who writes a language in
> forth? I don't. Most don't. Each has a view of what forth is or
> should be, but now they're all different. With nothing holding
> it together perhaps what we're seeing is the natural dissolution,
> entropy.
I see lots of writing languages in Forth ... mostly, of course, they
are one off languages, or languages that are turning existing data
into an executable source, like Jenny Brien's JenX parser ...
.... and since they have different goals than the goal of a macro
defining language, they don't have to have semantics as free standing
as a properly sandboxed Macro language ...
.... but that's a big part of the appeal of Forth. It gives the tools
to write a computer language in which it is easy to solve the problem
at hand.
|
