Hello Brian,
Thanks for the brilliant and depthgoing explanation of contexts.
I must admit that I was wrong in implying that word lookup somehow happened hieracally
at runtime.
Can we agree that word lookup is one-level (so to speak), but that you can still build
heiracal (someone *please* spell this for me :-) contexts. Even though this is done at
bind-time rather than evalution time?
I can see that this hierachy doesn't exist in a parent/child sort of way, but rather
by binding the words in nested blocks multiple times (at each context creation).
This is only happens with litteral blocks.
Let's try:
Giving 'a a value in the global context
## a: 1
== 1
## blk: [a]
== [a]
the lit-word 'a in 'blk is now bound to the global context
## reduce blk
== [1]
## o1: make object! [a: 2 ba: blk]
the words 'a and 'ba is created in o1's context.
The lit-word 'blk is first bound to the global context, later evaluated to [a]
But this evaluation doesn't occur until the binding-fase is over and thus the 'a in
[a] is still bound to the global context.
## o2: make object! [a: 2 ba: [a]]
In this case 'a in [a] is "found" at the time of binding (no evaluation is needed
since it's a litteral block), and thus bound to o2's context (because 'a has allready
been found in o2's context)
## reduce o1/ba
== [1]
## reduce o2/ba
== [2]
Seems to be true, please tell me if it's not.
Here's another funny thing that I've been playing around with:
(Fresh REBOL console)
## a: "global a"
== "global a"
'a now has a value in the global context
## proto: [a: 1 b: [a] c: bind/copy [a] 'self]
== [a: 1 b: [a] c: bind/copy [a] 'self]
the two a's in [] are now bound to the global context.
## print [reduce proto/4 reduce proto/7]
global a global a
making an object! based on the prototype
## obj1: make object! proto
changing the prototype
## proto/2: 2
== [a: 2 b: [a] c: bind/copy [a] 'self]
making another object based on the modified prototype
## obj2: make object! proto
Examining the binding of [a] in the objects:
## reduce obj1/b
== [2]
## reduce obj2/b
== [2]
How did that happen, did obj/a change?
## print mold obj1
make object! [
a: 1
b: [a]
c: [a]
]
## print mold obj2
make object! [
a: 2
b: [a]
c: [a]
]
nope...
What about the binding of the 'a in 'c then
## reduce obj1/c
== [1]
## reduce obj2/c
== [2]
hmm.
Best regards
Thomas Jensen
