Open Source

[Beathan]

As SL becomes more open source, I would like to explore the possibility of transferring and tranforming the CDS to its own servers, independent of Linden Lab. However, this should be a long term potential goal. I would recommend setting up an exploratory committee to evaluate the costs, benefits, pros and cons of such a move.

Any thoughts?

Beathan

[Jon Seattle]

Unless LL actually releases their own server code as open source, this will take quite a while. There are several small open-source projects to create new SL servers (ones that will work with the SL client) and grids but having tried some of those, I must admit they are really not very far along.

I will be very happy when projects like the CDS can do that though. In particular Linden Labs made certain decisions about how server resources are to be allocated that are worth revisiting. Are there alternatives to the "prim scarcity" economy of the SL grid?

There are other projects: Open Croquet for example, that may well come up with alternative 3D environments after some time. Right now Croquet is much much more programmable than SL, but the rendering is more basic, and there is a huge amount of work to be done on avatars and physics that makes SL a much better user experience. That may change in a year or two.

If LL does release their server code, it would be a great idea to set up independent servers. With today's hardware it would not be difficult to host tens of islands at a reasonable price.

[Beathan]

Jon --

You are exactly right. The LL open source project is in its infancy. However, I take LL at its word that it is moving towards not just open sourcing, but also linking open source sims to the world.

Thus, while we should not try to jump on this, we should start planning for it now. That is, we should have techies like you and FR and Luda brainstorm the technical side of things -- and beginning initial pricing of a server farm, and set up a budget to set aside funds for the eventual set up and operation of such a farm.

Beathan

[Bromo Ivory]

Try the Central Grid. The promise is amazing, even if it is definitely an "Alpha"

[Gwyneth Llewelyn]

The two OpenSim-based projects I've tried (Central Grid and OpenLifeGrid) are still at very early stages — about at the same stage LL was in late 2001/early 2002, although they're quickly covering the gap. There is no permissions system and no economy, and no way to transfer objects between avatars (except full perms). Scripting is limited, but curiously it is already implemented on top of Mono and in some cases with better results than LL's own solution.

I'd expect 1-2 years of development until it becomes a "usable" product at some time. In fact, what I really think is that LL very likely will never open source their own servers, but simply "take over" the OpenSim project and use it for their own purposes in the future. It's not as silly as it sounds — after all, the first practical web server, was created by the NCSA, and shortly afterwards, the Apache project reverse-engineered it to provide a better system using a more interesting licensing model. These days, of course, NCSA uses Apache for all their web servers

The secret, alas, is in the asset servers — which LL never considered to open source (not even IBM has access to them!). But it's true that they're very, very serious about interconnecting grids. Each time that it looks like LL might not release the open source code of their simulator software, they come out very clearly with the message that they fully endorse interconnection between grids. This is their major point right now — once they refuse to interconnect the grids, they'll lose control over the SL protocol, since the other OpenSim-based grids will soon find a way to interconnect to each other without LL's assistance. Right now, they're all expecting LL to allow access to their asset servers

[Beathan]

Yikes --

That could be fatal.

Beathan