> Elizabeth D Rather forth.com> writes:
>
>
>

>> Aleksej Saushev wrote:

>>> George Hubert yahoo.co.uk> writes:

>

>>>> On 10 Oct, 17:22, Aleksej Saushev wrote:

>>>>> George Hubert yahoo.co.uk> writes:

>>>>>> On 10 Oct, 12:25, Aleksej Saushev wrote:

>>>>> I would like to emphasize this. It is more important to deal with
>>>>> continuations than with naming and crappy standards of the past.

>>>> Then why bang on about vocabularies?

>

>>> They were mentioned by original poster for his reason, which is lost now.

>

>>>>>>>>>> But I don't remember the FIG-FORTH vocabulary
>>>>>>>>>> as being a spaghetti stack, rather a simple linked list.

>>>>>>>>> It is not.

>>>>>> Vocabularies in Fig Forth were simple linked lists where the initial
>>>>>> link is chained to the context vocabulary (or more precisely the link in a
>>>>>> dummy name structure in the parameter field), except for the base (FORTH)
>>>>>> vocabulary, which terminates the list with a null.

>>>>> I think we don't want to discuss how it is done in FIG FORTH,
>>>>> since the source is available, anyone can fetch it and read it
>>>>> with his own eyes.

>

>> The best reason not to waste time discussing how FIG Forth
>> implemented its dictionary links is that it's irrelevant.

>
> It is relevant, if you stay on topic which comes from Scheme land
> instead of driving it to discuss some irrelevant things like that
> of FIG model's drawbacks.
>

>> Although it may have been the first Forth many people saw, it
>> was not the first Forth used, and it's of only historical
>> interest now. The Forths in current use are all based on
>> different models; this is true not only of ANS Forth but also
>> the most interesting and popular variants.

>
> The subjects says plain, "The fundamental concept of continuations."
> Thus historical interest, and FIG Forth matters much more than all
> "micro", "mini", "poly" and ANS together because of its de facto nature
> combined with public domain source, which is easy to grasp.
>
>
>

>>>>> And I'm pretty sure that we speak about the same, the difference
>>>>> is you don't stress additional pointers, which are saved not in
>>>>> link fields but still influence overall logical structure.

>>>> What additional pointers; a linked list always has a head as well
>>>> as links (otherwise you'd never find the start) unless it's a
>>>> circular list (i.e. Round Robin, for co-operative multi-tasking).

>

>>> When you have links between its elements, it's not just a list now.
>>> E.g. if you have _any_ NFA above FORTH, you can get NFA of _any_
>>> other word in FIG-FORTH's initial dictionary. Do you still think
>>> it's just a list with no loops? As for me, it's obvious, that
>>> there are loops from TASK till FORTH and in for each
>>> vocabulary V, from V till V's LATEST.

>

>> Sorry, I don't follow that paragraph at all. Most Forths
>> (including FIG) link from the most recent word in the current
>> vocabulary to less-recent words. You can search more than one
>> vocabulary, but always in that order, and there are no links
>> into the middle of other chains. My understanding of 'spaghetti
>> stacks' is that there are either reverse links or links into
>> arbitrary points in other stacks. As no Forth dictionary
>> structure that I'm aware of is either a stack or supports links
>> into the middle of other vocabularies.

>
> My understanding that neither current standard nor ancient FIG Forth
> do not provide such "spaghetti stack" vocabulary. I explained above
> why it is so for FIG Forth: because of reverse link from FORTH to TASK
> at least. It is quite different from ancient LISP, which does
> not contain such loops (it really can be done).
>
>
>

>> ...

>>>>> In FIG-FORTH, brought as example above, you can FORGET any word
>>>>> above FENCE. That gives right to speak words are stacked.
>>>>> In X3.215-1994 you have the same semantics but not in CORE and
>>>>> marked as obsolescent.

>>>> You mean they're stored contiguously in memory. They are not a stack
>>>> in the sense you can push and pop items.

>

>>>> Most Forths use separate contiguous spaces

>

>>> We don't talk about "most Forthes", we talk about: a)
>>> continuations; b) one particular FIG Forth as called for
>>> instance by previous poster.

>

>> As I read the Wikipedia article on continuations, neither FIG
>> Forth nor any other Forth that I'm aware of supports them
>> directly (although one can construct almost anything, if one
>> wants to).

>
> Nothing surprising for collapsing language.
>

>>>>> And this conversation has started in Scheme part of world.

>>>> And cross posted to comp.lang.forth. If you simply wanted to
>>>> post incorrect information about Forth to Scheme programmers
>>>> why cross post?

>

>>> What is incorrect?

>

>> Virtually all your assertions about FIG Forth.

>
> Sorry? Just name them and explain why.
>
> Or is it just a personal attack meant to continue flame war?
>
> I'm pretty sure about interiors of FIG Forth, the code is in
> public domain, just point the place in the code and in words.
> Isn't it easy? Why two old-timers waste time explaining such
> easy things in abstract words involving "most Forths" and
> "virtually all"?
>

>>> Please, don't talk about any "most" or other things, FIG Forth is fixed.

>

>> Yes, as an obsolete architecture FIG Forth is fixed. But that
>> architecture did not support either continuations or spaghetti
>> stacks as described in the referenced Wikipedia articles.

>
> That's my point,
>

>> In any
>> case, I'm not sure how an obsolete Forth implementation is of
>> interest to Scheme programmers.

>
> The traces of that were deleted not long ago.
>
> One advantage of Scheme community is that the problem was quickly
> resolved, not to speak of Forth community, where it's involved
> abstract "most" and "virtually" just to attempt to prove someone
> is wrong with quite definite things like internal stucture of
> FIG-Forth vocabulary instead of solving problems originally set
> and reflected in subject.