World: r3wp

Yup. There is a lot to learn there, and the number one lesson is 
this: they did it with what was there already, no plugins. If we 
want REBOL in web browsers, we are going to have to look somewhere 
other than Google for inspiration. Still, some of their recent behavior 
has some clues. Look at Gears - they are clearly recognizing that 
networks aren't reliable or fast enough to count on consistency. 
That's why they are promoting local storage.

Perhaps it's the language

How do languages like Erlang cope with unreliable networks?

How so?

(sorry, question delayed by AltME)

Isn't Erlang designed with network failures assumed?

In general, languages like Erlang handle unreliable networks with 
redundancy, and it can do that because it is a functional language 
with no assignment. The state needed to answer a question is passed 
with the question. That way you can ask the same question multiple 
times and get the same answer every time.

Still, Erlang's advantages are more from the OTP system than the 
Erlang language, though the no-assignment language with lightweight 
process concurrency and message passing makes it easier to implement 
something like OTP.

REBOL's statefulness and heavyweight processes make this kind of 
redundancy more difficult, but it can be done.

I recall Maarten was going to port an Erlang database to Rebol ...

this database was distributed across the network or something ... 
don't recall all the details

Since he hasn't done it ... I guess these things are more difficult

They are, but to be fair he may be busy.

there are so many people now working on qtask .. it should be finished 
by now!

Qtask is never going to be finished ... it is imo turning into ratgher 
complex product ...

Maarten's stack was based upon Chord implementation. But he mentioned 
even more interesting stuff. IIRC, he also said it is already running, 
and once proven, he will release it.

Google is never going to be finished - it is turning into a rather 
complex product. No, it is a company or a platform, not a product.

Which ties up all the Rebol talent into one big hole

Qtask could probably stand to adopt Google's 20% time. I'll ask them.

20% time?

qtask is Rebol's Hadron collider

All google employees are asked to work on a personal project 20% 
of the time they work there. That means 1 day a week.

Interesting .. and who owns the IP?

Qtask is also the largest scale project REBOL has. There is a good 
bet that PITL innovations will be coming to REBOL from Qtask.

information does not escape from a black hole except as hawking radiation

I don't know who owns the IP, but it seems that many current Google 
projects stem from such 20% projects

I've seen nothing come out of qtask of benefit to the larger community

That may be the case with black holes, but not necessarily the case 
with Qtask. Insight from work on Qtask informs the people who are 
working on implementing REBOL. Plus, the long term goal is to open 
source Qtask itself (or so I've heard).

Nonetheless, we should be having this discussion in the !Qtask group.

Maarten has some good stuff done, but hasn't released it yet, as 
it's not *quite* complete.


Erlang/OTP is heavy on supervision trees, so things get retried if 
they fail.

Graham, sometimes you don't want to see, and sometimes you don't 
want to wait.

I am finishing S3 as we speak (this week) and I hope we'll release 
it to the community as open source. Done right though - visible in 
the right places etc.

Could be.

Chord: the real testing got stalled, I hope to return to it. I managed 
to express it in terms of a few functions. So if you can define those 
in REBOL (I did them using Rugby), and my implementation works... 
you can rebuild from the top down.

Graham, there is no "could" here. We'll do it right or not at all.

Right:

- documented
- PR
- into multiple commmunities (e.g. the AWS S3)

etc.

Maarten, I was replying to Gabriele.

Oh, well, you're welcome ;-)

Nice to hear that S3 is progressing.

Maarten what is status on sphinx scheme?

Thanks all for the "browser" input.

Will: done and working. Needs to be documented.

I want to pull a couple threads together for comment - the REBOL 
browser concept and the cloud/services parts such as S3.  I think 
there is great potential in delivering an environment that can bring 
rich network applications to the desktop but that support development 
without the limitations of the web browser/html/ajax technologies. 
 It can be less of a "browser" and more of a UI to services platform 
in my view.  Trying to shoehorn it into a web browser just because 
it is common is a mistake I think.

At the same time I do worry that a grand plan such as a REBOL browser 
is putting the cart before the horse, R3 as the base needs to be 
done and delivering on its functionality promises first.

Rod, I wouldn't  worry about a Rebol browser taking priority over 
the base technology. There is no mention of a Rebol browser in Carl's 
blog,  the  Rebol 3 blog or Docbase.  If  there is such a thing it 
is only being discussed in private which suggests that it is only 
at the "idea" stage.


In fact , I get the feeling that when people talk about the Rebol 
browser, thay are actually talking about the replacement for VID 
in Rebol3.

That last part is what I was picking up too - that we are waiting 
on the new VID but the design for it seems to be reaching for the 
REBOL browser?

No one truly knows what Carl is planning

I think that people are calling the VID replacement a Rebol browser...I 
don't think they are referring to an HMTL/CSS browser written in 
Rebol or even a browser plugin.

Anyway, Graham is absolutely correct.