World: r3wp

@Izkata Ithought that I smelled a lambda - maybe a la Rebol - but 
close enough :)

REBOL has a stack, like most programming languages, but no explicit 
manipulation of it. DO function! just evaluates the arguments. One 
of the many gifts of using an interpreted language.

@Brian So Rebol is not a stack-based language like Forth, or concatenative 
languages, like Joy, Cat etc?

Nope. I haven't used the others but like Cat a lot. Too bad the author 
has (temporarily?) abandoned the project.

@Brian OK! I can get rid of _that_ mindset in trying to understand 
Rebol :)

REBOL is more like an interpreted Lisp, with a Forth-like direct 
binding model.

@Brian kinda like newLisp, then - which I have dabbled in, and liked

newLisp is more compiled than REBOL, and has dynamic binding, which 
is a completely different concept.

Unfortunately, it is a bit hard to explain direct binding to people 
already familiar with lexical or dynamic binding.

@Brian I hear you! Please don't try ;) now now. I'm still in the 
KISS mode as far as Rebol is concerned

BTW, KISS refering to _me_  NOT you

We all have our moments :)

In the simple phase of your REBOL education, all you need to know 
is that we have something called "direct binding", where words have 
direct references to their values, but that in most cases it acts 
like lexical binding because we fake it. So write your code as if 
it's lexically bound until you run into problems, then come back 
and be new again for the next level :)

Cool! :)

I run into these next levels pretty often, usually after a conversation 
with Carl or Ladislav. It's no problem to be new again :)

This looks simple enough to be readable just finely:

http://blog.revolucent.net/2009/07/deep-rebol-bindology.html

One thing that attracted my attention in the article:


...interesting fact about REBOL blocks: By default, their evaluation 
is deferred.

 - not being a native speaker, I do not know, whether it means the 
 same as:


...interesting fact about REBOL blocks: They are not evaluated (i.e. 
understood as 'data', not as 'code'), unless an evaluation is explicitly 
requested.

(but that is just nitpicking from me, the text is fine)

It's that terminology bug all over again. Good point, and good article.

Well, about binding in Rebol,  it's not that hard to understand.
The context of any word is a hidden property.
Meaning it can be changed at any time.

It is very hard to find the best terminology to describe the properties 
(like "evaluation"), exactly because they differ from the common 
cases in other languages.

yes to "The context of any word is a hidden property.", but "it can 
be changed at any time" is a bit complicated by the fact, that it 
is "immutable", meaning, that you create a new word with a different 
context, when you want it, instead of changing the original

if its an internal (silent) reconstruction we don't bother.
the behavior acts like a change.
a: [obj]
bind a context [obj: 1]
do a
== 1

From my point a view the serie A has bit been modified.

*From my point a view the serie A has not been modified.

And the same word OBJ got a new context

Ladislav, I see your point, except it's not  a problem from my point 
of view. ;-)

Probably we shoold invent new vocables.
instead of "direct"  binding like Brian said.
I would say  "space" binding.
in the block: 
b: [ a a a a ]
Each word 'A' may have different contexts.
Beause they occupy diffenrt locations in "space"
That"s all. :-)

Funny :)

to go further.
If we admit that words may change of context during time.
Then, Rebol process "spacetime" binding

Strangely enough, they don't really change binding, they are replaced 
with new words of the same spelling and new binding. Words are immutable, 
like integers. That is what Ladislav was talking about. The functionality 
is great, but terminology that implies that words can change binding 
is not really true.

Yeah I see your point but I still can pretend.

It's not a spacetime law violation

After new binding,

Even if you say that the "cell" contains (hidden)pointers to different 
locations in memory.

The cell himself remains at the same location in the block and has 
the same "public" name.
So that I can say :

the same "cell" in the same block with the same name has a different 
context.

Yup.

But the interesting thing is that the immutability of words makes 
BIND behave differently when passed a word (which it can't modify) 
versus a block (which it can). So when you bind a block, you aren't 
modifying the words in the block, you are modifying the block itself. 
This is an important distinction that we shouldn't gloss over because 
that tends to confuse newbies later.

that tends to confuse -> *glossing over the distinction* tends to 
confuse

Yes, the main problem is, that the "From my point a view the serie 
A has not been modified." will shoot you in the foot, as demonstrated 
in the bindology article

yes finally, maybe I raved too much :-)

We all reach our limits there too :)

Therefore, the introduction of "modifies" comments is needed (and 
useful!)

Instead of spacetime modification, it's spacetime travel

hefty...

I used to use the term "applicative binding order" for REBOL's binding 
model, but later on Carl started calling it "definitional binding", 
which is a bit less descriptive but sounds better.

sounds better than spacetime travel ?

Sounds better than "applicative binding order". Time will tell if 
the increased explanation needed will be outweighed by having a cooler 
term.

I am a fan of the "definitional binding", since it is short enough 
(just two words, i.e. the same length as alternatives, like "dynamic 
binding" or "lexical scoping"), as well as different enough from 
the alternatives

It is cooler, yes :)

but what does that mean exactly ? Definitional ?

I mean, in that context :-)

I perceive it as referring to the fact, that e.g. the binding of 
the function locals is done when the function is defined.