World: r3wp

so what ? you didn't declare 'crap in the context.
SET doesn't modify the context.

>> global: 1
== 1
>> do test: funct [ ][ set 'global 2 ]
== 2
>> global
== 2

so what ? you didn't declare 'crap in the context.
SET doesn't modify the context the words.

I'm saying that if you use the set method, then your words appear 
in the global context which you may not want

So what ? you don't declare 'crap in the context.
I repeat: SET doesn't modify the context of the words.

Here  it's local.
>> c: context [crap: 0 set [crap] [1]] c/crape
== 1

Sorry for the echo, my connection is bad currenlty

It's Altme ....

I think Graham simply wants the block to be set within the context 
with as little additional code as possible.

If you want to use the set method of whole sale settting variables 
you have to declare them first in the context to prevent them from 
becoming global.  So what gain is there?

unless you do something like this

use fields [
	set fields data
]

this creates globals though

 - this is actually wrong, SET does neither "create", nor "globals"

And in R3, it will bypass the setting of locals with funct

 - in R2, as well as in R3, when using the SET function, you bypass 
 the collection of object locals. In R3, with FUNCT, you may not bypass 
 the collection as follows:

set [a: b:] [1 2]

neat trick

because 'funct scans the body of the function looking for set-words

Actually, I already used the trick in R3

with collect-words, an handy function, not well known

native by the way

COLLECT-WORDS is a native, but it is quite easy to define it in R2 
as a mezzanine

but not soo fast indeed :-)