World: r3wp

So if you have two worlds running on two seperate computers (both 
on port 5400) and you later wish to run them on hte same computer 
you are going to have a problem....

AS safeworlds expects both to be on 5400 and your going to need to 
assign one to another port...

That would be my understanding.  The only way I know of getting around 
that is by creating a new world with a new name on the computer hosting 
the two (while the first is running remember), then copy the world 
files from the one computer to the new world and it should recognize 
the files as belonging to that new world.  I don't believe the contents 
of the world has any reference to the world name, so the files can 
be moved from one world to another without issue.

I ran into the trouble of a world becoming unavailable.  I simply 
created a new world, copied the world files over, and all was back 
in order in the new world name... except I had to click on all groups 
to make the unread after synching to the new world.

that would all be so much clearer in a diagram  :-)

A note on the resync bug.......I just accidently let my laptop run 
out of power while online. Under Windows, that sort of simulates 
a hard crash.
When I rebooted, AltME had lost my user name and password.
I surprised myself by remembering both.


Then AltME set about resyncing about 20 groups. In some of them I 
can now see messages dating back to 2005....Apologies if you'd been 
expecting replies from me.


So, yet another method that may trigger a resync if you have groups 
that appear stuck :-)

Another recipe how to get a resync: wait for a storm (quite frequent 
here these days) and a blackout while running AltMe. As opposed to 
Sunanda's experince I did not lose the password, just got a resync 
dating back to about 2005.

so randomly trashing the db causes a resync?  no more dot spam needed??

We need to put this on rebolweek :)

There are several methods that partially work -- this is just the 
latest discovery. Not sure how effective it is, nor what side-effects 
there may be.

Where's Thor ?

I'm getting timeouts here again .. and yesterday it was down for 
a long period.

What would it cost to get this world hosted by safeworlds?

Slicehost is only $20 a month ...

I saw some poor responses and/or short outages yesterday too. Looking 
better today.

on an off currently...

I keep getting disconnected.  No one interested in a professional 
host??

possibly the world has outgrown altme's BE capacity/architecture. 
 to much data to handle.. causing timeout's in client conections, 
while its busy keeping tabs on itself.

If anything traffic volumes are down in recent months:
    http://www.rebol.org/aga-display-stats.r?world=r3wp&year=2009

(figures show [web-public] groups only. It is possible private groups 
are way, way up).

well. being an account of 10% of posts and not having posted for 
2 months... I guess I am a sizeable part of the decrease... but its 
summer, vacation time, and many of us actually have less questions 
with R3.0 than its earlier releases...  what I find strange is the 
decrease in the number of people posting...

The recent decline in non REBOL related concent, seemingly comming 
from a common accord that we should make the world more REBOL  centric... 
will have made a big difference too.

Looks like Carl's son is working on Altme2 http://twitter.com/Sean_SAS
8)

Fix altme1 first!

I'm loosing lots of posts lately.. 8(

Will -- one of the biggest AltME annoyances is that it throws away 
what you typed when it has to re-establish a server connection.

That can happen if you have dropouts on your online connection, or 
if the server is busy.

Those of us who have been plagued with the problem have learned it 
takes three keypressed to send any AltME post:
  ctrl+A ;; select all
  ctrl+c ;; copy post to clipboard
  ctrl+s ;; attempt to post

Getting kicked off multiple times today losing posts.

Has this world reached its limit?

the same here. Not sure about limits, maybe server problems once 
again?

R3-alpha is okay.  Are they both on the same server?

Booted again.

Anyone interested in building an FOSS altme alternative?

I'm thinking of something that has both a VID/Ext interface and a 
browser interface

Will be done in R3. R3 Chat system is said to use similarities to 
AltME system, just more robust. It just needs GUI client ...

The biggest failing with Altme is the lack of a distributed server.

That will also be an issue with the R3 chat.

If we don't make sure we have a robust backend .. it doesn't matter 
what the front looks like, we will hit issues.

The other issue is the need to maintain a continuous tcp connection 
to the altme server.  That is a major source of issues.

not sure how Robust R3 chat is. After all - it is just one of possible 
architectures. It is true that I doubt it is much advanced. It is 
good for Chat, even threading, moving threads here or there, ranking, 
etc., but we are imo not following latest developments of distributed 
apps. Some few years ago I read something about Cademlia etc. protocols, 
but not sure what it is all about. OTOH - look at Altme - it is running 
for years. And if its back-end would be DB based, it would be enough 
for a long time probably.

I saw a similar issue with playsite when I used to play games online. 
 Kept on being booted due to crappy internet connections or

Maybe we can shift to using Amazon's SDB .. they allow you to store 
millions of records for no charge

Messages can be sent using the SQS once someone has committed a new 
message to the store.

This gives us a backend, search, threading, no centralised server 
requirements, no particular person needing to host a server.

You can build a front end using a browser client or a View client.

And you can use https if you wish too.

I thought about using an IRC server but again this is a single point 
of failure.

Too ambitious??

Unable to connect for the last hour ...

I haven't been on consistently for a while, so I haven't noticed 
outages, but nothing is too ambitious for you Graham. :-)

ambition, and the ability to fufill are two different matters :(

The only issue that I can see is that there is apparently no anonymous 
access to SDB.