[REBOL] Re: Context - code included- 2nd version
From: g:santilli:tiscalinet:it at: 14-Sep-2001 22:13
Hello Ladislav!
On 13-Set-01, you wrote:
Some cents from me...
LM> You are contradicting yourself above, aren't you? In the
LM> first sentence you are telling that there are no "Special
LM> Words" and in the second one you are telling, that you call
LM> them "words not bound into a context".
It depends on the meaning you give to the word "special". If I
read "Special Word" I think of a word that is different from
Normal Words
; your special word is, instead, a word with its
context attribute set to null (or none, if you prefer :).
LM> ("words not bound into a context") exist, the collection of
LM> such words exists too. The only point you can make is, that
The fact that you can collect something does not mean that the
collection exists. You could collect all the gold in the world,
but does the collection of all the gold in the world exist? Again,
it depends on the meaning of the word "exist". For an explanation
to be easily understandable, words should be used with their
common meaning, or it should be stated that they are being used
with some different meaning.
Your "Special Context" has no attributes or properties. I don't
find it useful, but this is of course only MHO.
LM> Context (to translate it to your terminology: has got a
LM> reference, that is the same as the reference that any other
LM> "word not bound to any context" has).
I see your point, but the notion of "no context" seems to me
simpler and more understandable (i.e. less confusing) than the
notion of "special context".
LM> This is a legitimate POV, of course. I am motivated by those,
LM> who don't think that the "replacement of the terms" is enough
LM> to be able to understand the rules of the game.
Of course you are right. But isn't it better to use terms that are
easier to understand and require less definitions and
explanations?
LM> Yes, the CFUNC-generated functions are performance-wise
LM> unacceptable for some uses. The performance of
LM> CFUNC-generated functions might become more satisfactory if
LM> it were implemented natively using a highly optimized
LM> algorithm. OTOH, in that case the performance might be even
LM> better than the performance of the functions generated by
LM> FUNC now.
I still tend to think that not having to do anything is faster
than having to do something (i.e. rebinding, or just creating a
new context). With the current implementation the programmer has
the choice.
Regards,
Gabriele.
--
Gabriele Santilli <[giesse--writeme--com]> - Amigan - REBOL programmer
Amiga Group Italia sez. L'Aquila -- http://www.amyresource.it/AGI/
