Horrible Server Lag Needs To Be Fixed ASAP

[Silverwalk]

They could fix it by:
1. Getting the client to send a position update when the player clears the red area.
2. The server calculates whether the player takes damage when it receives the first position report after the cast completes.

[SKYeXile]

They could fix it by:
1. Getting the client to send a position update when the player clears the red area.
2. The server calculates whether the player takes damage when it receives the first position report after the cast completes.

Server prediction using client timestamps could fix any current issues with lag and movement.

[Silverwalk]

You're right but sadly for whatever reason they are not doing that - using a simple vector would be good enough.

If the update period is 0.3 seconds and the cast time is over a second the server should have at least two updates showing the player moving.

Draw a line between these two points and extend it out by the time until the cast ends, viola! Now you know where the player is at the end of cast with enough accuracy to solve this problem.

Probably would put too much load on their server software I'm guessing.

[SKYeXile]

You're right but sadly for whatever reason they are not doing that - using a simple vector would be good enough.

If the update period is 0.3 seconds and the cast time is over a second the server should have at least two updates showing the player moving.

Draw a line between these two points and extend it out by the time until the cast ends, viola! Now you know where the player is at the end of cast with enough accuracy to solve this problem.

Probably would put too much load on their server software I'm guessing.

im not sure it would, Planetside 2 current uses serverside hit prediction at a much higher tick rate than 3, i imagine the strain on that would be much more taxing than just movement in this, but then I imagine SE are running servers from 1987.

[TheRac25]

Server prediction using client timestamps could fix any current issues with lag and movement.

this is impractical from a computational standpoint and often times inaccurate

what needs to happen is:
1. clever tricks such as, the client should report its pos when it enters/leaves a danger areas as some people have allready suggested
2. the 500+ms internal server latency needs to be addressed, horrible, i couldnt even design such a system if i wanted to without using sleep functions. feels like they are directing traffic through a bunch of low end linux machines with ipnat and realtek network cards
3. they have race conditions that need to be fixed, plenty of screen shots showing people healing themselves for more than the next damage they take yet still dieing