animats

Well, yes. Extra round trip delay will do that. Add a full second and double region crossings fail almost every time. Region crossings extrapolate movement during the stall which occurs at every region crossing. Region crossings require a round trip handshake with the viewer (the underlying cause of many problems), so adding network latency and you will slow down region crossings. Are you running a current version of Firestorm? Firestorm does, in fact, limit bogus rotations at region crossings to 20 degrees and position error to about half a meter. Technical discussion here. Without that, you'd sometimes roll much further or momentarily go underwater, if you were pitching or rolling or moving vertically going into the crossing. On "smooth water", you should get smooth crossings. Should be better on AWS sims, because the region crossings are faster. That's a common problem with vehicle scripts. You have to check that you still have animation permissions after a region crossing, and not do anything avatar-related until the avatar arrives in the new sim. When you get the CHANGED_REGION event, the vehicle has crossed but the avatar(s) are usually still catching up. A recent fix made objects cross regions faster, and now it's more likely that you see avatar delay. So scripts have to wait for the avatars. The way to check is to call llGetObjectDetails on the avatar's key and ask for its root prim. If that's the same as the vehicle's root prim, the avatar is properly seated. To make this less frequent, LL recently put in a delay so that scripts don't restart after a crossing until either the avatars have arrived or something like half a second has elapsed. This makes that problem less frequent, but it really should be fixed in scripts. It's important that scripts restart after a crossing even if the avatars haven't arrived, because during that period you can't steer but may still be moving. Scripts need to take appropriate actions, like stopping. This is more of a problem for land vehicles, where you're not far from going off-road or hitting something. There's one curve in Satori near GTFO HQ with a double region crossing, and many vehicles go off the road there. There's also a jog on Robin Loop like that, where region crossing problems with naive vehicles regularly result in landing in someone's yard. Fortunately the resident there isn't in much. You have to manage region crossings in LSL, annoying though this is. Really, though, with most crossings now down to 500ms or so, the viewer and scripts can mostly cover for the delay and you barely see it.

Anyone tried using DynamoDB on AWS? It's a key/value store. The free tier offers 25 GB of data storage 2.5 million stream read requests per month from DynamoDB Streams 1 GB of data transfer out per month, aggregated across AWS services but writes are US$1.25 per million writes. Amazon does all the backups, it's a redundant system, etc. If SL is up, it's unlikely to be down. It uses HTTP requests in JSON and responds in JSON, so you should be able to access it directly from LSL. It's similar to the experience key/value system. It would be useful if LL offered this as an add-on for SL accounts, so you didn't have to set up an account with AWS directly and it was authenticated through your SL account.

Here's a detailed look at the wardrobe from Thoroughly Modern Millie. This is a 1960s interpretation of 1920s styles. A Thoroughly Modern Wardrobe

Flooding has been simulated that way. There are pictures on here somewhere.

Especially since problems like that, where sim-side state is messed up, are entirely server-side.

The web browser inside the viewer may not be able to play whatever type of video you're trying to play. (What's really going on with media on a prim is that each prim showing a web site is launching its very own web browser. The browser inside is Google Chromium, an older version and limited in some ways. You can't play DRM content, for example. By the way, the process that runs the browser is called "Dullihan", for some reason.)

I looked into using Rasa and TensorFlow so my NPCs could have semi-intelligent conversations. It's a big job, though, and the usage is too low. There are chatbot programs that improve and can be improved if talked to enough, but they need a lot of conversations for training. Nothing in SL gets enough traffic for that.

There's a free set of Slink alphas the creator gives out. They work on some other bodies, too.

Luca, the leader of the militant furries, has the last word on this. She's often in SL and is a good resource for this question.

On a related note, I have a JIRA in, BUG 229442, to make linked doors play well with pathfinding. Right now, if you give a building pathfinding attributes, all linked doors will become immovable and scripts that try to move them will generate script errors. With that fix, doors would still move. You'd have to use each door once, then rebuild the navmesh, after which your building would be pathfinding ready. Buildings could even be prepped at the factory and would be pathfinding-ready on rez. This would be useful for my NPCs, Virtual Kennel Club dogs and cats, and a few other things that use the navmesh.

If it's at the Akamai level, whether you see overload throttling depends on where you are and to which cache server you're connected. Akamai's world of caches. This is what makes it possible for everyone to view the home page of CNN or Fox at the same time. It only helps for SL if more than one Second Life user happens to be both in the same area of SL and the same area in RL. Or some of the people in the same area in RL are seeing some of the same items in SL.

Really, mostly tedium. There are 7 sizes and three packages, a new version is a pain to package, and I was off doing something else when their IM came in via email. I think you're right, though. I should make them mod, but put in the resizing checker.

"No user-serviceable parts inside." It's an interesting question. I sell my escalators as no-mod because someone tried to resize one just a little, which broke it because the moving parts have to fit for it to work. Then they were unhappy. Then I had someone else who was unhappy because they wanted a mod version to re-texture, and I didn't want to bother with a custom job. What do creators with complicated objects find works best?

That's rotation around a single axis. Any object rotating freely without outside forces has some axis of rotation. The axis will not change without some outside force. This is why gyroscopes work. Rotational velocity is described by a vector. The direction of the vector is the axis of rotation, and the magnitude of the vector is the speed. That's what you feed into llTargetOmega. Changing the axis of rotation by applying an external force is called "precession". There are good articles and videos on line about how that works.

Yes. The history is that the US was formed as a voluntary alliance. Version 1, the Articles of Confederation, was set up more like the United Nations, with a very weak central government. Version 2, the US Constitution, had a much stronger central government. The US federal government is weaker than the central government of France or England, where Paris and London run things. It's closer to Germany and Switzerland, where there's some regional autonomy. The effect has been to over-weight rural areas. This has some interesting effects. US interstate highway system. All states have at least one, even the ones with few people. All mainland states except Maine, in the upper right, have at least two. That's what it took to get Congress to pay for it. UK motorway system. Only the heavily populated areas are served. Strong central government.

The Unreal Engine tools don't really do visual content creation. You create in Blender or Maya and import into Unreal Engine, where there are some tools for optimizing content and assembling content created elsewhere. (I'm not a UE4 expert; I installed it once, built some demos, and tried a few things.)

The 7 types of VR users Someone's take on the available demographics for virtual worlds. The "VRChat Bro" is included. Ask yourself which of those groups might be drawn to SL.

Right. The viewer is happy, system load is just the viewer and is low. System memory usage is about 60% on the graph. The viewer is just sitting there waiting for assets to arrive. Viewer is the current Firestorm beta, the one with EEP. Clicking "edit" on the no-texture areas doesn't bring them up. That normally increases the priority of showing an object, so if it's loaded but not in the GPU due to GPU memory space problems, it gets brought in when you edit it. But there's no texture info the GPU can find. Speculation: Akamai, which runs the caching system used for SL assets, has detection for overloads, and will throttle requests from an IP address if some threshold is exceeded. Maybe that's it. Akamai's system is tuned to deliver web pages. Teleporting somewhere in SL looks to Akamai like loading a web page with a thousand images or so. That's an abnormal event. A quote from Akamai: "An online video game download consumes as much traffic as about 30,000 web pages, so that's why Akamai is working with companies such as Microsoft and Sony to reduce global Internet traffic during hours of unusually high demand." Due to the US elections, we're probably in a period of unusually high demand, with news pages constantly changing and being reloaded. I've brought this up at Server User Group, and it's on the list of things to look at after "cloud uplift".

It took 5 minutes to load this scene. Most of the time the network was almost idle. 45 Mb/s is available. Texture loading was very, very slow today. I went to New Babbage and buildings were showing with opaque windows for the first two minutes while textures loaded. Above is Robin Loop, which I drive around every few days. This time, the road texture never showed as I drove the 13-region loop. Equally bad with cloud and non-cloud regions. Are others seeing this?

Larson Logistics' quarry train and the streecar from Sumi at Burns seem to use newer scripts that work well. I've run the quarry train the length of the Bellessaria line without problems other than those from defective signals. "Model Railroading is Fun!" - old issues of Model Railroader

Take a look at the trains in Jane's Viking sim, Folkvang. Those are all keyframe animated. So are the trams in New Babbage, I think. Keyframe animation will work well in overloaded sims, which is why I use it for my NPCs. Those appear to move like avatars, but they're really keyframe animation precalculated dynamically a few seconds ahead of motion. That's a good way to run trains, which follow predictable paths along tracks. Trying to do it by constant feedback is in theory better, but with script overload the new normal, precomputation is the way to go.

I checked my logging, and on some region crossings there's a meaningless "CHANGED_LINK" event.

I have one, "animatsalt", for testing. Non-premium, L$ balance 0, no inventory, with a good male minor-brand free mesh avatar from The Free Dove. Used mostly when I need a vehicle passenger or a second point of view.

Once LL gets the provisioning automated, private regions should be limited only by the ability of people to pay for them. AWS capacity is very easy to add.

OK. "Pack" is a function in Python for making binary structures. ">BLBL" is an instruction to the Python packer on how to pack up that data as raw bytes. I'm amazed that they got far enough to send a UDP datagram to a sim correctly with so little code. What's going on here is that you did the XMLRPC call to the login server. It then replies, and tells you the address info for the sim running the region into which you are logging. From that point forward, most of the communication with the sim uses binary UDP packets. These are encoded in a format described here.The actual binary format of blocks is defined by a huge machine-readable text file called "message_template.msg", which you can find in the viewer sources. Decoding this format is Not Fun. Each UDP packet contains multiple SL messages. The messages are just run together; there's no structure that lets you pick out just the ones you want without decoding all of them. There's Lua code to do this in the Open Simulator world, and C++ code inside the viewer. Probably C# code in Open Simulator. SL has other communication protocols. There's a serialization protocol called LLSD. It comes in two flavors, XML and binary. It's self-describing, like JSON and Google protocol buffers. That is, the names and types of the fields are in the data. Once upon a time, around 2009, there were apparently people interested in this level of detail about the system. That was before my time, but you can go back and find info. There was an Architecture Working Group, and Project Snowglobe, and much interest in building the Metaverse. There's a monument to Project Snowglobe in Hippotropolis. Go visit and read the posters. That totally ended by 2008. But the documentation is still around, there's Open Simulator, and there are third party viewers.