Thinking Rebol: WAS {Re: Re: New Rugby based chat client}
[1/4] from: jasonic:panix at: 24-Sep-2001 12:02
> We Europeans are simply too far from New Zealand in provider terms (at
least
> I am). Loading the Reblet takes lots of time as well, as is logging in.
Even
> when you have async behaviour this will be annoying because of the delay
> between messages.
>
> Would be interesting to host the server in Europe as well and see what
> happens.
hmm.. interesting.
What is the precise delay.?
How did you track it [traceroute]?
What kind of arrangement of servers would make this fly?
It seems that Rebol and the indicated IOS applications must all depend to
some degree on addressing these issues. Designing applications that tend to
just-in-time instead of real-time or near-real-time.
I thought that Rebel was created partly to provide a low-profile platform
for implementing all this.
But the impression I am getting is that the same basic problems exist. ie
local Client processor performance now far outstrips network latency issues,
and heavyweight servers seem to be an essential part of the solution.
Am I reading this all wrong? I hope so...
My hope is that good topology and fast local applets which combine
server+client ubiquitously can work brilliantly. At least when
communications are among relatively small groups of people sending
relatively small amounts of data and meta-data. ClientServer combos can
maximize use of metadata. I believe much net communication is among small
groups of people most of the time, often only a handful or less at any time.
This is the nature of real workflow.
I hope all this does not sound to vague. I am a multimedia develop not a
network engineer, but my observations are mostly based on r&d design
consulting for past 5 years with European businesses. Most projects involved
clients who expressed a need to develop collaborative business workspaces
over internet. But usually with they had little or no hands-on awareness of
the web or internet habits.
Just addressing workflow and getting them past the widely publicized
paradigms of internet = THE WEB + email was first problem. Second problem
was lack of toolkits for rapid prototyping of Internet based applications so
we could all start playing with their problem to focus on the real workflow
needs. The CEOs and project managers almost all argued they wanted
well-defined heavy top-down solutions before even had even really
experienced themselves hands on what we were talking about. Years of
expensive software development contracted out to database consultants had
convinced them that nothing small or cheap or non-MS could be much good..
Rebol would seem to be an ideal platform for rapid prototyping, and its
longer-term follow up implementation. Rebol/View offers a wonderful brevity
to smart module desktop net apps.
I am asking everyone here for broad design wisdom for using Rebol in
distributed applications.
[vaguely, but sincerely :-)]
What the limitations are..?
What the advantages are..?
What would a Rebster look like?
Do we have the pieces now ?
If it we want to use Rebol for sharing graphics, calendars, whiteboards and
PDFs among restricted, trusted groups of users [<25 say, typically 5 -10
people] what would you recommend for system specs and architecture?
Is it still way too early to have this discussion.. come back in 2002 after
much more trial'n'success?
When can we publish the book: "Thinking Rebol"
hope this makes some sense..
thanks
- Jason
[2/4] from: m:koopmans2:chello:nl at: 24-Sep-2001 18:29
You are partly right. You need a toplogy of servers, so that clients can
connect nearby. And for New zealand is far away (on the Reb on the Web on
the ICMP)
--Maarten
[3/4] from: jseq:mediaone at: 24-Sep-2001 12:44
Jason,
I suspect that /Express points to REBOL's solution to your problem. It's a
push server kind of like Marimba - REBOL applets and data are synchronized
on the clients machine at some scheduled interval.
Push is a pretty good caching strategy for the kind of small applets/data
combo that Reblets will function best on, but /Express is neither here
today or free. If you're looking for something like that, you might want
to consider an approach with an existing peer-to-peer framework that
possesses synchronization capabilities - distributing Reblet/Data bundles
using something like Groove et al. is something I've played around with,
and it works pretty well. There are other frameworks, but Groove is the
most polished I've seen firsthand and has a pretty low learning curve.
[4/4] from: jasonic::panix::com at: 25-Sep-2001 12:13
John
Thanks for your comments.
Yes I remember the Groove announcements. Have not had a chance to play
around with it yet.
I will look into it some more. I have been looking to do these kind of
things using Zope and or standalone Python modules
www.zope.org
www.python.org
Both of which led me to Rebol. I clearly need to educate myself better now
on some of the lower level network topics. Rebol can play very well with
Python/Zope systems and I trust both those communities to offer and support
affordable well engineered options.
One example is the ZODB, the persistent database used by default in Zope,
but which is now being explored by the larger Python community.
http://www.amk.ca/zodb/
Another to review is ZEO [Zope Enterprise Objects]
http://www.zope.org/Products/ZEO
..as you can see, I am still trying to figure out what Rebol does best, what
the parallels are to other toolkits/platforms and where it is heading..
- Jason
----- Original Message -----
From: "John Sequeira" <[jseq--mediaone--net]>
To: <[rebol-list--rebol--com]>
Sent: Monday, September 24, 2001 9:44 AM
Subject: [REBOL] Re: Thinking Rebol: WAS {Re: Re: New Rugby based chat
client}