World: r3wp

How many scripts did you try to port?

I have no plans on jumping into R3 at this point, since there are 
so much even on R2 I need to learn. But for the future reference, 
is there any plan for a tool (or mode in Rebol itself) to help Flagging 
out those R2->3 differences... For example, I remember seeing that 
PICK works differently on R3 (right, unlike R2 which is offsetted 
by one), it'll be quite hard to spot all those from source alone, 
since parameters are often defined at runtime?


Some changes will of course be obvious (for spotting), like sockets, 
since their parameters have been canged a lot. but difference on 
data readiding/writing (ascii/binary/unicode etc) may hide itself 
quite a while.

Hmm I need to read some things again to be sure I have even got it 
right..  (and I'm sure this is issue that has been wel discussed. 
before, sorry)

Yes, there is a difference in Pick behaviour, but keep in mind, that 
behaviour is different only in exceptional cases

Ladislav? Really? So just go and try rebol.org submissions, one by 
one, and then tell me, how many of them actually will work ...

I did not tell that porting would be difficult, but I better try 
to tell ppl, that simple do %my-r2-script.r in R3 might not work 
right away ...

But I have no problem standing corrected :-)

Wrong, it is you who stated something without trying

Even if such a statement was correct, I insist, that mine is correct 
too: uninformed and unsourced opinion.l

Be nice, boys :)

Well.. The fact of life is that *if* you can find an computer system 
(even officially stadartized programming language) that works exactly 
the same 10 (or more) years, it's development obviously must have 
ended..  100% backward compatibility is totally impossible, question 
is how well the situation is handled, documented etc..

There's no comparison draft; if it's what you're asking

Well, OK, may have been too harsh, sorry, I just don't like unsourced 
statements like the above.

A good idea that, especially once we get out of alpha and things 
stop changing so much. Most of the core semantic changes have gone 
through already - current development focus is in areas with no corresponding 
concepts in R2. We'll see how much the multitasking affects things, 
and the proposed object! changes.

the time invested in learning R2 is not lost when you switch to R3.

Just avoid to pass to much time on GUI aspects and Ports handling, 
they are/will-be completly redisigned.

Plus there is a project to backport some of the R3 improvements to 
R2 so you can use some of the fun new toys. A project I should get 
back to soon, I might add. Some things won't be compatible but it's 
amazing how much can be made to work.

must be cause R2 is Turing complete

Naw, that's not it. That wouldn't explain the stuff that can't be 
easily backported without rewriting REBOL from scratch.

Joanna asked about the forward compatibility: R2->R3. There is quite 
decent backward compatibility (R3->R2), as BrianH is proving, and, 
of course, the forward compatibility is even better.

Steeve: well, sometimes things take time..  It took me decade (or 
was it two) before I got <build  does> .more or less right.  And 
I still don't get Lisp.. (nor have I tried it in years)..

mostly, speed-up improvements. except for 2 functions AFAIK.
Do you know which ones guys ?
(Brian, you don't play :-)

It might be easier to get this than it is to get Lisp. Lisp says 
that code is data, but it isn't necessarily so. Code realli *is* 
data in REBOL, at runtime, and the whole language is bilt around 
it. Once you get that it's amazing how easy the rest gets.

What I mean about lacking compatibility: Things like ports are totally 
different, so we can't directly port protocols. They have to be rewritten. 
VID is also an entirely new system that bears little resemblance 
to the original VID. AFAIK using DLLs is also very different. But 
of course many functions may work the same, even if they are rewritten 
underneath.

I bet, that it would be easy to port the old VID from R2, to R3, 
but it is unlikely, that anybody would want to do it.

yes, it might be possible. there were also talks about emulating 
VID at some point. maybe that's better left for a hobby project.

Brian: I noticed that I had earlier (ears ago) missed the Rebol word 
  Compose  entirely..   No wonder some things seemd to be ackward..

'Compose was an addition to get round some awkwardness ...

i'm not sure it's easy. The flow of GUI events is drived very differently. 
It's could be a burden to simulate the R2 behavior.

Well, Compose is not comfortable. More comfortable replacements were 
proposed at http://www.rebol.net/wiki/Replacement.

To be honest Ladislav, i don't see the interset of the INLINE function.
What can be done with INLINE, we can't to currently ?

The answer to your question is dead simple: nothing.

ahah, a lighter syntax ?

In fact, the Inline function can be implemented as a mezzanine, which 
would serve as a proof, that you can do without it, if you wish.

But, the same applies to the Compose function, it can be implemented 
as a mezzanine too.

Or even Reduce, if you have Do/next.

As a half.funny sidenote. I was reading http://www.rebol.net/wiki/Parse_Project
and there is comment about people having CompSci degrees.. I have 
one, and I still feel a bit dumb ...

You can certainly ask, if something is unclear to you in that page.

nah.. I think I just need to stretch my brains a bit and try to remember 
those things I did learn at the Uni years ago..

Not sure it helps

Right, some things may be newer, than what you learned at the university

And I''ve never been so much on theory side..  I've preferered practical 
things..

Well, the attempt was to define the dialect to be as practical as 
possible, which certainly isn't an easy task.

As a practical ... I ment things like Embedded,  measurement  & control 
with some asm+c code, self-made wireless protocols etc.. Of those 
I know something (and got  my master´s degree).

The parsing model of R3 is based on a theory that didn't exist when 
I was last in college :)

To make you feel better: there are at least 20 proposals in the article 
I personally don't mind about (alternatives, that weren't chosen, 
as well as some keywords I do not plan to use).

Re the practicality: a good Parse dialect may be very practical (can 
be used to implement dialects, parse texts, match various patterns, 
etc.), but, at the same time, the design of it is rather a theoretical 
task.

Which is happily over :)

Yes, although there is a possibility, that a new requirement/need 
appears.

But, anyway, I like the result so far.

The most recent Parse finding is probably CureCode #1401, which serves 
as a proof, that While is more "universal/fundamental" than Any (or 
Same), which may be found "crippled" in some situations, just because 
it "knows better than the user what to do" - I personally hate such 
software and am happy we convinced Carl to at least introduce the 
While keyword/operator.