[REBOL] Re: URL & setword bugs
From: gscottjones:mchsi at: 18-Aug-2002 5:54
Hi, Paul,
From: "Paul Tretter"
> The first part you referenced is not a problem.
> We often set multiple values this way - such as:
>
> this: that: none
>
> This way everything can be set to none.
>
> Gotta be careful with this though as
> some datatypes have problems doing this.
>
> And for your second example the test:i would be getting the value at index
> 'i and not setting do to the placement of the ":".
>
> try:
>
> change test/:i "BBBB"
...
does give ...
>> test
== ["a" "BBBB" "c" "d"]
and then
>> change test/:i "b"
== "BBB"
>> test
== ["a" "bBBB" "c" "d"]
which may or may not have been then intended action. I ask myself which
result is the *least* surprise, and my guess is that it works correctly as
is. So it seems that
change test/:i "whatever"
is an equivalent action to
test/:i/1
but my expectation of Andrew's proposal would be
test: ["a" "b" "c" "d"]
i: 2
test/:i: "BBBB"
test ; pseudo results ----> ["a" "BBBB" "c" "d"]
I personally would like to see this type of action; I think it would add to
the "intuitive" nature of the language. It is not clear that this behavior
would break some other action/result.
Can anyone else see that it does? Just curious.
Thanks, Paul.
--Scott Jones