Second Life News Update Week 32

I’m a little behind getting a Week 32 news update out today. But, other things to play with and write about.

Metareality’s Gianna is busy with something, so there has been no podcast for the last 2 weeks. The podcasts were our major source of first hand news on the Mesh Deformer. So, we have no clue where we are with this project.

Main Channel Roll Out

We have had lots of misinformation about this roll out and a load of misunderstanding. Fridays meeting of the Server/Scripting group got some more information out on the updates.

Server-Scripting UG Meeting

Andrew Linden tells there are a couple of things that crash regions running the new PathFinding/Havok 2012.1 software.  Falcon Linden has fixes that are being added to the server maintenance package. Those should be on an RC this coming Wednesday.

The two problems are rare, random, and not easily exploited by griefers or so it seems.

He also commented about a problem with Traffic Count saying: “Meanwhile, there is a mysterious drop in parcel traffic calculations that doesn’t look real but we haven’t figured out why they aren’t showing up right. It is a systematic drop that is not reflected in any of the other data.

In any case, parcel traffic is not used for search ranking anymore, from what I hear.  Yeah, it is overrated so no need to pay attention or worry about it.”

Oops! I keep telling you the Lindens only know their particular part of the system and even less about how we use it.

He continued, “I’m currently pulling a bunch of logs so I can try to calculate the amount of traffic at the data “source” to either duplicate the traffic reductions across multiple days, or not… Whatever the outcome it will give me a clue as to where the problem lays.

Andrew also made it clear that the deployed Pathfinding is running in production mode and the debug code has been removed.


Andrew had this to say about the camera:

Oh, that reminds me of something I wanted to bring up here… for the most part the server just accepts the camera position as the viewer declares it. But, not the avatar position… the server controls that.

However… the server’s interest list has a hybrid strategy… sometimes the code uses the avatar’s position to figure out what is in view, and sometimes the avatar’s [camera] position.

So… while I was doing a minor cleanup pass over the interest list, in search of ways it could be fixed, I considered fixing this weirdness to just use the camera position for figuring out what is in view. However, there may be some worries about such a strategy from the residents. So, I wanted to get your opinions on the matter.

Jonathan Yap brought up what was expected to be a prime concern: If your camera is far away and an avatar appears near you then your radar might not register that fact.

Andrew sees another issue as: “The other issue… a minor technical one, would be that the camera can move much faster than the avatar under normal conditions, which might make for poorer sorting (incorrect, slower) when the camera is moving a lot, which might lead to a little viewer lag, depending on implementations.

On the other hand, a virtual fly-trough with the camera would be more correct.

Group Editing

Baker Linden gave an update on his work:

 On Tuesday, I successfully displayed a group of 20k members in the viewer — some data was missing, but the data loaded. However, there remains an issue with the sql query causing issues with the server. 🙁

I have now found about 4 solutions on how not to fix the management of large groups. My next move is to switch over to use a cap, serving the data through HTTP. Thankfully, the time spent learning how this part of the server worked and the code I just wrote has to be ported over to utilize the cap, so it’s coming along pretty well (apart from the fact that I’m rubbish at writing Python).

Oz (Linden) announced yesterday at the TPV meeting that I’ll be talking about this at the next TPV meeting in two weeks. The data will be sent in a new format (hopefully reducing its size) and TPV’s will need those specs.

That’s all the news I have.

It looks like the editing fix is still weeks away. But, it is moving closer.

Leave a Reply

Your email address will not be published. Required fields are marked *