[Opensim-dev] Guest logins for grids, an idea which could change a lot

Wed Dec 17 13:41:55 UTC 2008

Truth is that was my vision as well somehow. The best way I had in mind to describe it was the forum and IRC examples, but yes a big reason why I was thinking about this would be that Opensim should as well be usable as a platform just like the web, which you can surf freely even where you don't register if allowed to. I think guest for grids would be crucial in allowing Opensim to look and work exactly like "3D web" system. This would be extremely useful for closed grids as well, when Opensim would be ran in an intranet network and require no account authentication and registration there.

About the Public Standalone mode you mentioned... I believe that the current standalone mode does work just like that. As far as I know, Standalone mode only means that a separate Opensim instance cannot connect and run their region to you from the outside, but you can run in standalone over the internet and allow someone else to log into your sim from the outside. So I think that such a thing is at least 85% implemented if that would be correct... anyway that's a separate thing from grid guests but indeed important as well.

My exact vision was just implementing the same "accounts_authenticate" flag used for standalone in opensim.ini.example for grid mode as well. Apart from that it would then be necessary that the last name is Guest and probably other checks as well, but that's kinda what I went with in my idea. I think that logging into a grid anonymously is something separate from teleporting from one grid to another with hypergrid, although a close concept. As for a modified viewer, the Hippo Opensim viewer is indeed the best one to rely on for that.

From: stefan at tribalmedia.se
To: opensim-dev at lists.berlios.de
Date: Wed, 17 Dec 2008 09:50:59 +0100
Subject: Re: [Opensim-dev] Guest logins for grids, an idea which could change a lot

Mircea,

When Darren and I drew up our visions for the 3D web, 2 years ago, we were already then seeing a 'mesh' of standalone 3D regions, used for everything from closed private desktop applications (logged onto with local authentication) over social worlds (where building identity is key) to public product demos and casual gaming (which could be accessed anonymously) - and the user would just hop as easily between them as between web pages in a browser.

This is why we concentrated so on the region, and for OpenSim to be more of a framework, not an implementation.

It is only recently the discussions about how the 'mesh' itself should be organized have started moving. The hypergrid was a big step forward to fulfilling the vision. We are now on the verge of being able to let people publish _regions_ not connected to any grid - self-sustained entities serving all that is needed.

These concepts of yours fall well within this vision - we need to extend on the authentication mechanisms to allow for these cases you mention, and the ones I added as a response.

Of course, as I understand it, much of this should probably be available today - if you set up your own grid, you can start hypergridding, your identity being under your own control (more or less)

So, I'd say the next steps would be:

* Getting a viewer modified to support 'hypergridding' by being able to specify a region url and mimicking a legacy-compatible hypergrid teleport out of that, so that the current legacy backends are kept synchronized. Maybe something for the Hippo crew?

* Introducing the 'public standalone' mode that has been discussed on this list, where all grid services are started up from within OpenSim.exe, but listenin to ip ports so that it can function as a home grid in itself - and making sure this mode is as easy to get running as 'private standalone'.

* Being able to 'connect grid to grid' in hypergrid - what I e-mailed about earlier, being able to register a whole grid onto another grid, so that you can start building conglomerates.

The 'public standalone' should not be a problem at all - the problem there lies, as has been discussed elsewhere, in maintaining codepaths.

The solution is, as always, to do careful refactoring. The current startup code is a mess, and there is no real division into 'proxy' and service - it's there, in the code, just not reflected in names or layers.

Best regards,
Stefan Andersson
Tribal Media AB

_________________________________________________________________
Discover the new Windows Vista
http://search.msn.com/results.aspx?q=windows+vista&mkt=en-US&form=QBRE
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://opensimulator.org/pipermail/opensim-dev/attachments/20081217/c2ce2ff1/attachment-0001.html>