[Solipsis-tech] RE : [Solipsis] some questions about Solipsis (from Taiwan)

[KELLER J. FTRD/DMI/I. <joa...@rd...>]

Hi Shun-Yun,
I am happy to know there is a solipsis-like project elsewhere in the =
planet :-) Sure I want to get your netgames 2004 paper !
Your questions:
Q1: A list of nodes, at least some nodes should be alive. To get that =
list you can download it, (eg entities.met in the solipsis distrib) =
and/or while using Solipsis you keep memory of met entities so the list =
is frequently updated.
Q2: By now: Checking is done for all known entities (whether they are =
inside or outside SOI); When entity A detects that an entity B has =
entered the SOI of entity C, entity A sends out immediatly a message to =
entity C. If entity B is outside SOI of entity C, nothing occurs.=20
Q3: So so. We have proved that in some tricky cases it will not work. =
But we have strong hints that it should works most of the time (not =
proved).=20
Q4: I am not sure to fully understand this question. There are some =
deterministic rules to choose entities that will restore or keep GC =
(something like "take the nearest that fulfill GC property"). But all =
that is probably not very predictable since unpredictable network delays =
have a huge impact on rules like "take the first node you got and..."
Q5: The awareness radius, awareness area or SOI is enlarged or reduced =
in order to keep the number of entities inside in a certain range (say =
between 9 and 15).=20
=20
Have fun,
-- Joaquin
=20
=20
BTW: we plan to handle multiple (human) languages in Solipsis GUI, and =
our first target (after english) is Chinese (simplified-mainland and/or =
classical-taiwanese) we will be happy if you and your friend contribute =
on that.=20

	-----Message d'origine-----
	De : Shun-Yun Hu [mailto:sy...@ya...]=20
	Envoy=E9 : mardi 8 juin 2004 15:53
	=C0 : KELLER Joaquin FTRD/DMI/ISS; SIMON Gwendal FTRD/DMI/ISS
	Objet : [Solipsis] some questions about Solipsis (from Taiwan)
=09
=09
=09
=09
	Hello Joaquin and Gwendal,

	I'm a CS graduate student in Taiwan, I've recently learned the very =
interesting and wonderful works you guys are doing at France Telecom =
R&D, on Solipsis.
	=20
	I got very interested in your work because I've also been thinking =
about a massive P2P Virtual Environment that exists on the Internet, and =
I've been doing related studies and research since Sept. 2003, along =
with a good friend of mine.=20
	=20
	I've only come across Solipsis recently, and had read through both of =
your papers. However, I've some questions raised in my mind. If you =
could answer these questions for me I'd very appreciated. :)=20
=09
	Q1. what's the bootstrap procedure? (i.e. how does a node join Solipsis =
initially? through a list of known nodes? or through a 'gateway server' =
of some sort?)
	=20
	Q2. when, and how frequent does an entity check for mutual neighbor =
discovery? (every time when someone within the Sphere of Influence (SOI) =
is moving? or at certain intervals? another question is about the range =
of checking. According to the paper, seems like checking is done only =
for those entities within your SOI, however, to ensure Global =
Conneectivity (GC), it seems like it's possible you would connect to =
entities outside of your SOI, then will you check for those entities =
whether they're enterining someone else's SOI?)=20
=09

	Q3. A very important question is, is GC guaranteed to be maintained if =
every node keeps the "inside convex hull" property? I understand you =
seem to be thinking that is the case, however, has this been proven by =
either simulation or mathematicial analysis? (I'm very curious to this =
question, because I've also met with a similiar problem).=20
=09
	Q4. To ensure Local Awareness and Global Connectivity, entities need to =
connect to each other, however, are connections between entities =
uniquely determined, given their locations in the virtual world, or are =
there actually more than one possibility of these 'links' between =
entities? (I'm thinking the case when an entity is trying to ensure GC, =
the entities connections it end up doing, might not be done in a =
strictly deterministic manner, or is it purely deterministic?)
=09
	Q5. under what situation would SOI be enlarged? (the paper describes =
how it could be reduced, when there are too many to connect, what about =
enlargement?)=20
	=20
	sorry for these long questions, I'd appreicate very much if you could =
answer them. :)
	=20
	We've got a paper that'll appear in the NetGame 2004 workshop coming up =
in early Sept. also on P2P networked virtual environment. If you're =
interested, I'd be willing to forward a copy to you.
	=20
	Thanks again and cheers! :)
	=20
	=20
	Shun-Yun Hu
	=20
	=20
	=20
	=20