New software rolled to the main and RC channels this week. There isn’t much to say about the rolls. We hit some problems with last week’s rolls. Fixes for those problems are in this week’s RC channels.
The main channel got the Large Object Rez upgrade that Baker Linden worked on. The idea was to place handling the rez tasks in separate threads. This allows the simulator to continue doing other tasks while a large link set is rezzed.
Blue Steel and Le Tigre are getting the same package. It fixes a crash mode. Presumably this is some exploit fix. I think that because almost nothing else is being said about it.
The Magnum channels got the package from last week with fixes for problem found last week. The known fixes are; region managers being exempted from Direct TP disable, neighboring region visibility issues, and bogus User Not Online error messages.
There are some crossing problems that are being worked on. Vehicles are having problems. Some are seeing problems walking across region boundaries. Some of those problems are likely Interest List related. So, Andrew Linden is looking at those problems and was working on a fix Tuesday.
Interest List Additions
There is a new addition to the Interest List change coming out of QA and going into an RC, likely next week. Adrew or Maestro will be talking about it at Thursday’s Beta Server meeting. The code is running in ADITI now, just on certain regions: Solariam (adult), ADITI.
ADITI does enforce Adult/PG settings. So, that can be a problem. Andrew is going to try to have a PG region setup by Thursday. They copy regions from AGNI to ADITI for testing when needed.
This change is purported to improve scene rendering. I suppose making it faster and smarter, with smarter meaning rendering things near you and not things far away. Andrew says if you clear your cache and go into the region, you should see a faster rez of the things near you.
The Solariam region has 1k+ objects in it on the ADITI side. The AGNI side version has about 9.8k objects. So, they are busy regions and a good test for comparing main grid code and the new code running only in ADITI.
Setting your Max Bandwidth will affect the render time. It seems that UDP currently controls that performance. Most of the developer types at the meeting set there Max Bandwidth at 1,500kb/sec. There is a UDP problem that starts around 1,900kb. The Firestorm Support team and Dev’s have recommended for some time that people set their Max Bandwidth below 1,900.
To really see the new code working, set your Max Bandwidth at 300kn/sec. Your rez will be slow, but you can more easily see the order in which things are brought in, which is the improved Interest List at work.
With HTTP Texture Get enabled textures will not be affected by the Max Bandwidth setting.
There are some odd behaviors with Pathfinding Characters, the mini-map, and visibility. If you are seeing things that just look wrong, like a PF Char moving in front of you on the mini-map and not being visible, file a JIRA. Again the Lindens will need as precise a time as possible.
Clear Visible Cache
With the Interest List changes some of the methods of clearing the Visible Cache has stopped working. Most of the methods that are now failing depend on moving the avatar to a high elevation then back to the previous location. This forced a couple of updates into the servers Interest List for your avatar.
To improve server and bandwidth performance some delay has been added to the Interest List. So, a ‘quick’ trip up and down is not registering as a change and an update is not recognized as nedded by the server.
The same problem is experienced when people change their Draw Distance (dd) to zero and then back to whatever, 128m. Previously this would force an update of the Interest List. But, no longer.
The fix is to move the avatar or change the dd and wait 5 second or so. Then the server believes the camera did actually move and updates the Interest List for your avatar. During that 5 seconds your camera needs to be still.
I used to be able to set dd to 0 and everything but my avatar and the ground would disappear. The server now enforces a minimum draw distance of 10m. So, everything may not disappear.
Rendering Things Behind You
Andrew explains: “The server sends what we’re now calling “cache probes” (ObjectUpdateCached messages) for the cacheable objects behind you and if they are in cache then the viewer will put them into the render pipeline, even if they are out of view, at least… the viewer puts them on the mini-map.
The other bit of work that is new in my pending project is that I’ve expanded the definition of what is “cacheable” for the object cache data. Currently it is the case that “cacheable” means it is static and doesn’t have any script.
The reason for this is that we’ve basically got two big lists of objects on the server: “passive” objects that don’t move and don’t have scripts and “active” objects that are moving or have scripts. The two lists were being used for two different purposes: (1) for script execution and (2) for interest list sorting cacheable vs non-cacheable.
I changed the interest list to look on both lists and send cache probes and updates for any object that hasn’t changed in a couple of minutes, which greatly expands the set of cacheable objects. So, there should be more cache hits going forward.”
Adjusting the Gray Goo Fence (GGF)
GGF is a set of limits to stop griefers. It limits how fast things can be rezzed and sets other limits. One of the limits apparently not covered is sound generation. That seems to be a growing problem. The Lindens are looking at whether they need to add sound limits as part of the GGF.