World: r3wp

Yes, it's still a word! and not a logic!, even if you reduce it.

So what turns it into a logic, if not reduce?

do

reduce the block instead

>> if ( do first find [false true] 'false ) [print "false is in the 
list"]
== none

looks like R3 does not exhibit that behavior. it's reduced to logic!

Ok, great, that's the case I was worried about then...

Or LOAD:
>> load to string! 'false
== false

I'm confused as to why reduce, which seems to perform other evaluations, 
would not (in R2) turn false into a logic!... Henrik, you are saying 
it does?

Fork, Rebol 3 does, so this may be an inconsistency in Rebol 2.

It's often works, if you turn stuff into a string and then load that. 
Other ways to turn stuff into strings:

>> load mold 'false
== false
>> load form 'false
== false

*It* often works ...

Not quite;    reduce 'false  is a word,  reduces as a word!     reduce 
[false]   reduces the block, evaluating false to the logic! false. 
  It's fun stuff.   ;)    mold/all  can be your friend exploring 
the ins and outs of evaluated values.

oops! :D

Sorry John; responding to Forks confused post not yours.   I always 
trust Geomol advice  :)

I got confused too. :)) REBOL is funny to play with!

>> type? load form 'false
== word!
>> type? load mold 'false
== word!

In R2, REDUCE main purpose is to reduce block! values. For general 
evaluation use DO.

btiffin, I make mistakes too! :-)

Yeah any word! reduced is still the word!

sorry:

>> type? false
== logic!
>> type? reduce false
== logic!

R2 and R3 behave identically here.

Henrik, yes, but
>> type? reduce 'false
== word!

Is that the same in R3?

It seems like reduce is a core notion, and it would be nice to see 
a list mapping what conversions it performs.  Changing that between 
R2 and R3 seems rather... fundamental?

Fork;   There are also "lexical forms" for all these values.   #[false] 
 is loaded by REBOL as the logic! value false.  No need to evaluate 
in that case.

John; yes.

I'd suggest coming up with a new keyword, e.g. reduce3, if something 
like that were to change.

geomol, yes, they are identical

Fork, we're normally ok with not being completely compatible with 
new versions. Better to correct the failures as soon as possible.

Brian;   It comes down to the ill-documented sequence of  load/evaluate 
and the spellings of some the values.  false, true, none are the 
most common gotchas.

Fork, R3 is not meant to be compatible with R2, so that would not 
be likely to happen. If a function doesn't behave consistently, it 
will change. Just for the record. :-)

Alternative:

>> type? do load form 'false
== logic!

But the most simple must be:
>> type? do 'false
== logic!

There are so many ways to get the same result.

Ok, I appreciate the help... but I do wonder if there is a certain 
answer that Rebol 3 reduces words to logic

So my earlier example, would it act different in R3?

>> if ( reduce first find [false true] 'false ) [print "false is 
in the list"]
false is in the list

Nope;  words reduce as words.

Here is a good article by one of heroes;   http://en.wikibooks.org/wiki/REBOL_Programming/Advanced/Interpreter

same result

and it would be because we are in that example always dealing with 
words.

Ok, so Henrik, does that mean you retract your statement "Fork, Rebol 
3 does, so this may be an inconsistency in Rebol 2."

yes :-) sorry about that. I mixed up some results.

In regards to what I said: "Im confused as to why reduce, which seems 
to perform other evaluations, would not (in R2) turn false into a 
logic!... Henrik, you are saying it does?"

Ok, whew.  Good :)

Give Ladislav's article a read...it cleared up a lot for me, but 
I still can't explain it  :)

reduce does not turn words into logic.  do does.  I'll write that 
down :)

Thanks for the link, I will check it out.  Ladislav seems to think 
about some of the same general issues I do, as he apparently made 
lfunc... we're probably on the same page in other ways...

Fork, you also have a problem with your check, if 'false isn't in 
the series:

>> first find [true something-else] 'false

** Script Error: first expected series argument of type: series pair 
event money date object port time tuple any-function library struct 
...

generally I only use REDUCE on blocks, because it does not seem very 
useful on anything else.

Yes, I knew that bit :)

ok

Geomol, that works in R3. :-)

Yep;  he seems to "get it".  Normal rebols just flail about until 
it sinks in.

I've programmed since 6502 assembly, have an EE degree, and have 
designed very abstract DSLs.  Nothing shocks me.  :)  I know all 
the ways one can make decisions, I merely need to understand the 
decision set that was made in REBOL.

the reason is that statistically it makes better sense to have FIRST, 
SECOND, etc. be gentler, so we don't have to check anything if we 
reach tail.