World: r3wp

(it's just too damn ugly :-))

Are some parts of R3 already considered "final" (minus bugs, the 
need to reconsider, ...) just like "for the moment we believe there 
won't be any major changes"? 
And if yes, are these documented somewhere?

I am not sure. Some kernel parts for sure. Just recently we await 
release of Unicodised R3, which internally changed many things. Then 
Unicode will be kind of final :-) Modules and plug-ins should follow.

Dunno, maybe networking kernel is final, we just need more schemes. 
Currently we have only http scheme.

I was specifically thinking about network part, and wether it would 
be worth digging in deeper.

I would guess that until unicode & modules & tasks are given some 
soak time, many many things are up for change.  Nothing in the area 
of mezzanines would be stamped final yet (imho) as Mr. Hawley hasn't 
optimized away  each and every (while few) stray bit yet  :)  (plus 
DevBase has yet to roll large).   Umm, in details ... I'm pretty 
sure Carl has nailed down most (or all) of the conceptual layers 
and the interfaces won't change as much as internals might.

Ingo;  I would say dig in.  If only to blaze trail for the rest of 
us.  :)

Hmm, but maybe networking will not be affected that much. It is already 
device based propably, running upon async core. I think that studying 
it, including http 1.1 Scheme from Gabrile Santilli laboratories, 
is worthy experience anyway :-)

While we're at it ... I just tried 

>> echo %http-scheme.r 
>> probe System/schemes/http

and got a "Rebol system Error #1405: I/O error"
 Program aborted abnormally ....

... I forgot ... this is on linux through wine, so maybe it's not 
really rebols fault. I'm not sure about the current when and wheres 
of error reporting for us "outsiders" ;-)

Confirmed with 2.99.4.3.1 4-Jan-2008/17:22 build native Win98

Umm, confirmed crash

Is it possible to get the source somewhere else?

write %http.r mold system/schemes/http

Thanks sqlab, I was actually stuck in my own thoughts here ...

Question on alternate UI options, specifically ChUI (terminal style) 
or mobile options would be of interest to me.  Is anything planned 
or expected that would apply to those areas of UI implementation?

it's not planned right now, unless someone steps in and does it, 
but it requires VID3 to be about done, before that can happen. The 
focus is currently on a scalable DRAW based user interface.

but it would be fun to do a low-color, low-res VID3 GUI. :-)

Thanks Henrik, some comment recently made me wonder if I had missed 
something on that front.  I like scalable DRAW based as an option 
too. *smile*

IMHO, this is the right way to go:
http://en.wikipedia.org/wiki/Dynamic_Language_Runtime
My hope is that REBOL3 will be implemented over this framework.

So do I, though not as the main implementation.

Why this one? Aren't there also other VMs, which could be interesting? 
LLVM, Parrot?

Ah, .Net related ... so integration ... well, we might need to support 
that one, of course if it allows us to interface .Net, other than 
that there is no sense .

prot-http.r should be in devbase. also... we'll eventually publish 
the docs.

That's why I like it, Pekr. It seems to me to be the best way to 
integrate .NET with REBOL using fully managed code.

Aside from that though, not much point. Even if you are talking about 
the Silverlight or Compact runtimes, it's still much larger than 
REBOL on its own engine. As for speed, who knows; a lot of smart 
people are working on the DLR.

As for REBOL3 on .net ported on the DLR I think of having available 
all the libraries, classes, and object of the framework and from 
other languages.

This will bring much attention to REBOL, a wider odience, a vast 
amount of material avaiable, the opportunity to write bigger applications. 
It is a whole new word that opens. Also, REBOLERs using of the Microsoft 
platform will migrate to the "pure" REBOL when needed.
I  consider this step a winning move from every side of the view.

Also, as other languages are ported to JAVA a porting of REBOL on 
it would open a lot of opportunities from this other side too. Look 
at this article for an overview of JRuby: http://www.javaworld.com/javaworld/jw-07-2006/jw-0717-ruby.html

Things such as JRuby and IronPython are independent implementations, 
so you are free to reimplement REBOL in Java and .Net...

Kaj, having the idea does not mean I am able to implement it !

Well, that was really my point. It may sound blunt, but having an 
idea does not mean that others have time to implement it

But at the same time this does not mean it is a bad idea ;-)

Tell me about it. I've been wanting to integrate REBOL with .NET 
since before the DLR came out, but no time yet :(

Brian, lets hope that REBOL Tech will have a project to port it on 
.NET using DLR. It would open to us a lot of doors. The whole microsoft 
.NET world is full of object and libraries which would be fantastic 
for us.

And, as written, even the pure REBOL would gain a consistent crowd 
of programmers enlarging by a magnitude its user base.

Somehow, I don't think porting to .NET and the DLR will be high on 
RT's priority list.

Hope is the last thing that dies....

Gregg, what you think about opening to the .NET world ?

Giuseppe - I think that any kind of possible integration has to be 
a good thing, no? :-)

If you want R3 on the DLR, pay someone qualified to implement it. 
It doesn't have to be RT - R3 is a community project.

Wait until the R3 Unicode semantics are settled though (real soon 
now).

first native C/C++ header import functions
to easy unlock the full power of fast, stable C/C++ lib's
second unstable, bloated COM support if really
necessary ;-)

If you are talking about Windows, you have to assume that C/C++ libs 
are unstable too, and that you can't do anything without COM.

On Vista you have to assume that there is a lot you can't do without 
.NET support - many of the new APIs are .NET-based.

C/C++ doesn't mean fast either - it depends on the algorithms.

for me it is enough if "nearly all" C libs have been
more stable and faster than COM container I have ever
used. if you are happy with COM, use it! I prefer C/C++
lib's if possible. at least even TASM is unstable and slow 
if your design is poor but this is splitting hairs...

The problem I have is that what I need to do with Windows has only 
COM APIs, or in some cases .NET. Unstable beats unavailable.

It depends on what RT's goal is, who has the time and skills to do 
things, and what's most important. I would like to  see time spent 
on an official systems for packaging, building, IPC, an IDE and other 
tools, a solid OSX build, etc., rather than a .NET/DLR port.

.NET sucks (from what I can see), but boy, what a market... (hopefully 
that one sentence sums it all up).

Community project, remember. Who said RT had to do it?