[sldev] Point of no return in teleporting?

[John Hurliman]

Teleporting in Second Life is something of a gamble. If the grid is 
healthy and concurrency is normal, your teleport will generally always 
succeed and things are fine. However at times like right now, you may or 
may not ever reach your destination, and you may or may not get 
disconnected in the process. I was looking in to the disconnected part 
with some bot code and found that in those cases the client receives 
TeleportStart, TeleportProgress, TeleportProgress, and the connection to 
the current simulator times out before the teleport fails or finishes. 
Is the current simulator assuming that all teleports succeed (until 
further notice) after TeleportStart and that's why it is disconnecting 
too early? The real problem here is too much backend load, but it seems 
like clients could deal with it a lot more gracefully.

Could the client detect if a timeout is imminent and send a 
TeleportCancel to revive the current sim connection? Or could it 
re-establish a connection to the current sim without having to login 
again? If so, what is the point of no return (either in length of time, 
specific TeleportProgress packet received, however it is measured) 
before these efforts will always fail?

John Hurliman