Go to content Go to navigation Go to search

Pathfinding Panic

August 8th, 2012 by Marianne McCann

Usually I spend a lot more time being thoughtful before I post here. It means my entries are usually few and far between but, I hope, worth the wait. With all the brouhaha about pathfinding today, I opted to hop right into it.

First, let me start with a critique for anyone from the Lab who might be reading this: this is why you need to communicate. I am doing your job here, but if you guys would do a better job at communicating to your own paying customers, people would get better information than what I’m about to lay down here. Please rethink your communications policies.

So today Pathfinding code went live on the main channel of the Second Life server software (the code that runs all the regions on the grid). Tomorrow it hits the LeTigre and BlueSteel Release Client server software. That will mean that every public region on the main grid will be running server software with pathfinding code, including a Havok physics update.

Pathfinding is something the lab has been hammering on for months now. Makes prims that can act sort of like critters. They can wander, attack, evade. Kinda cool, if you’re into that. I’m not so much, but who knows. Maybe I’ve just not see the “use case” that I can use.

Meanwhile, a blog for a third party viewer went out, claiming that the pathfinding code will use up to 18% of your sim resources, and provided information on how to either optimize or disable pathfinding for your region.  This, in turn, led to what one might call “the usual panic.” Instantly, it became fact that “the evil ol’ lab was out there hurting its users, and it was up to all of us to fight back my immediately disable the code.”

The usual talking points were trotted out. The lab has clearly not tested the code. It is deliberately trying to “kill” SL by rolling this out, etc., ad naseum.

The thing is, this code did not just pop up on the main grid today. It was used in The Wilderness project and Hairy Hippo Fun Land months ago. It was sprinkled around the grid then as well as a “PF” release client. Five weeks ago, it was promoted to the Magnum release client server software, and has been chugging away ever since. Just ask anyone who uses mesh vehicles, and they’ll tell you all about the fun of trying to cross out of a Magnum viewer and dealing with the mismatch in Havok versions.

But what you won’t hear much about is how their regions were crippled by a staggering 18% of sim resources being tied up in pathfinding. The regions themselves have been running fine, albeit there have been a few bugs along the way — and there are still a few things that need to be ironed out. Some issues with the way some prims travel on the navmesh, for example.

Now, to be fair here, the code can use some resources. It is designed to throttle itself, however. It does not use all the resources all of the time, and is intended to deliberately not take resources from other uses. It is not going to be crippling your region.

So really, what I want to say is this. Enough with the knee-jerk reactions. I understand that the lab has not always thought things though, and was even fairly antagonistic to their userbase during the “M” era. Their current lack of clear global communication, too, doesn’t help to get people beyond that time.

But rather than seeing a blog post and instantly reacting to shut it all off, do your own tests. Go visit other regions that are running the code (which by the time you read this may well be all of them). Heck, try your own region on for size, and see if its running poorly compared to, say, this time last week. Don’t take the word of any blogger — even me. Go look with your own eyes, and form a rational opinion of your own.