World: r3wp

Is there a reason why codecs (like image loading) are not in host 
part?

Oldes - AFAIK, codecs are going to be completly overhauled. We wanted 
streaming support, and current implementation is imo rather primitive. 
Carl agreed in one roadmap document release, but that file is gone 
(we are waiting for new one). I hope proper port based API will be 
available. So - on one hand it is a good thing it is not part of 
the host-kit, as we arelly need one standardised API, not myriad 
of different hack-ins, but otoh Carl could benefit from some community 
experiments in that regards ....

Just curious: Is there a maximum length to a word name? can you TO-WORD 
a 1 GB string, assuming it's valid?

easily checked with a little powers of two loop ;-0

don't have the memory :-)

have a break :-)

I'm writing docs. Just getting warmed up. :-)

Max word lenth is 255 or 256 in R3

ok, thanks. that's fair enough.

http://www.curecode.org/rebol3/ticket.rsp?id=580&cursor=1

Btw Henrik, I read your GUI R3 code. It's R2 fashioned. You missed 
some R3 tricks. I suppose you already know it though ;-)

Steeve, depends on which parts you read.

I give some examples.

The 'merge-values mezz is actually like the native 'resolve

you use a lot the old idiom
select block 'word 
vs
block/:word (which is faster and more compact)

(sorry it's: select block word)

it's quite possible that some functions are from the original R3 
GUI, long before newer tricks were involved.

Another old R2 idiom:
all [in object 'property object/property]
vs
select object 'property

var: either cond [...][none]
vs
var: if cond [...]

hmm... yes, that is old

But you know, We all are in the same trend, It will take a while 
before we loose old habits

Today, I saw something horrific in an old script of mine.
>> 8 * to-integer v1 + v2 / 8

V1 and V2 are actually integers

(not related specificaly to R3)

I let you think about this coding horror ;-)

Was that R1 semantics? ;-)

I wouldn't always use the IF version in place of EITHER. I generally 
prefer to be explicit.

We use that IF or UNLESS returning none trick a lot. It is used so 
much that it might as well be considered explicit. It is good to 
remember that REBOL doesn't have an optimizer; we have to optimize 
by hand, and the IF trick is faster than the explicit EITHER.

It is used so much that it might as well be considered explicit.

As is my preference for avoiding premature optimization. ;-)

On that note: Steeve...

>> a: [a 1]
== [a 1]
>> a/:b
** Script error: :b has no value
>> select a 'b
== none

It's the same thing as the SELECT object word trick.

Gregg, don't avoid optimization altogether in the name of avoiding 
"premature" optimization :)

A better aphorism to use here would be to avoid being "penny wise 
but pound foolish" - it makes your point better :)

Agreed on optimization being required at times. Not loading lots 
of data? Yes. Using a hash instead of a block? Yes. I have even avoided 
the mezz loop constructs at times, though more often for clarity 
than performance. 


Maybe I'm a special case, but I have rarely found the need to optimize 
my REBOL code for performance at this level. And I can safely say 
I have never ever, ever optimized out EITHER for IF. Since you said 
it's used a lot, what code should I look at to see how and why? I'm 
always happy to see what I've been missing.

It's used in mezzanines, for instance. I tend to just write code 
in micro-optimized form in the first place, since it is easy to do 
and saves you the trouble of doing it later. Even if you are macro-optimizing 
the code (refactoring and such) you tend to use the same micro-optimizations 
in the new code. For that matter, many of the changes and enhancements 
in R3 were done to make micro-optimized code cleaner to read and 
write than they are in R2. But I mostly write library and mezzannine 
code nowadays, so micro-optimization has greater impact than it would 
for user-level code.

Two notes:

- I optimize for code size, using IF, or UNLESS

- the average speed-up looks like being about 27%, though, which 
is worth considering

Interesting: That's the same speedup that referencing words from 
object contexts gives you, relative to function contexts.

Agreed: Less code means less code to maintain and debug :)

I wonder, what is the purpose of the --- function, did not suceed 
to find that out.

>> same? :--- :comment
== true

aha, thanks

I hope, though, that everyone will use the word COMMENT.

I expect that Carl added it to be able to write more concise code. 
I prefer COMMENT too though.

I tend to just write code in micro-optimized form in the first place

And do you consider that premature optimization, or not?

It is just as easy in theory to write the code in micro-optimized 
form as it is to not, so why not? In practice though, in R2 it is 
harder to write micro-optimized code (for various reasons). In R3 
it is *easier* to write micro-optimized code than it is to not, because 
we have been deliberately been optimizing the language for just that 
effect for a couple years now.

in R2 I use none of the mezz loops(they really slow down code), now 
that they're all native, I have much more options in R3.

What's more, we have been working to make the behavior of the language 
much more consistent between datatypes, so it is now easier to make 
macro-optimizations like changing datatypes without needing to change 
the code that uses them. This is why the SELECT trick for objects 
works the same as it does for blocks, for instance.

Pity I can't use any of it, because I still need to run on Cheyenne

I ran into the same problem, which is why I made R2/Forward.