[DiscordArchive] I know you were involved a little bit with the discussion before, what came up last time with walkli
[DiscordArchive] I know you were involved a little bit with the discussion before, what came up last time with walkli
Original source
I know you were involved a little bit with the discussion before, what came up last time with walkline's implementation that this PR does differently?
rektbyfaith
09-23-2025, 03:00 PM #1Archived author: Bench • Posted: 2025-09-23T15:00:45.893000+00:00
Original source
I know you were involved a little bit with the discussion before, what came up last time with walkline's implementation that this PR does differently?
Original source
I mean his implementation before it was merged, one where (iirc) sphere intersection was used
rektbyfaith
09-23-2025, 03:01 PM #2Archived author: Bench • Posted: 2025-09-23T15:01:09.804000+00:00
Original source
I mean his implementation before it was merged, one where (iirc) sphere intersection was used
Original source
I don't fully remember walklines implementation, but I do vaguely remember those discussions and I believe he started doing this, and then found a simpler fix for his specific bug instead.
rektbyfaith
09-23-2025, 03:02 PM #3Archived author: Pursche • Posted: 2025-09-23T15:02:36.941000+00:00
Original source
I don't fully remember walklines implementation, but I do vaguely remember those discussions and I believe he started doing this, and then found a simpler fix for his specific bug instead.
Original source
Well fingers crossed this one goes a little smoother then
rektbyfaith
09-23-2025, 03:03 PM #4Archived author: Bench • Posted: 2025-09-23T15:03:37.439000+00:00
Original source
Well fingers crossed this one goes a little smoother then
Original source
Here: https://discord.com/channels/21758927576...8996179978
I believe the outcome was that the +0.5f magic offset moved from being applied on the result, to being applied on the start position of the ray being used to figure out the height under player, in his specific bug.
rektbyfaith
09-23-2025, 03:05 PM #5Archived author: Pursche • Posted: 2025-09-23T15:05:41.927000+00:00
Original source
Here: https://discord.com/channels/21758927576...8996179978
I believe the outcome was that the +0.5f magic offset moved from being applied on the result, to being applied on the start position of the ray being used to figure out the height under player, in his specific bug.
Original source
Which may have solved his specific bug, but kept the underlying issue that collision testing is done with completely different collider "shapes" (only the client actually uses a shape while the server uses a simple position with no width), so if the server uses such a collision test before moving the player to a position, it's possible that the server finds a good position to stand on which the client immediately falls through because it resolves the collision COMPLETELY differently.
rektbyfaith
09-23-2025, 03:07 PM #6Archived author: Pursche • Posted: 2025-09-23T15:07:46.129000+00:00
Original source
Which may have solved his specific bug, but kept the underlying issue that collision testing is done with completely different collider "shapes" (only the client actually uses a shape while the server uses a simple position with no width), so if the server uses such a collision test before moving the player to a position, it's possible that the server finds a good position to stand on which the client immediately falls through because it resolves the collision COMPLETELY differently.
Original source
Now a sphere test might not even be the most accurate way of doing this, but it should at least error on the side of pushing the player further away from problematic spots.
rektbyfaith
09-23-2025, 03:09 PM #7Archived author: Pursche • Posted: 2025-09-23T15:09:53.871000+00:00
Original source
Now a sphere test might not even be the most accurate way of doing this, but it should at least error on the side of pushing the player further away from problematic spots.
Original source
And it's gonna be way cheaper than the upside down pyramid that the client actually uses.
rektbyfaith
09-23-2025, 03:10 PM #8Archived author: Pursche • Posted: 2025-09-23T15:10:10.318000+00:00
Original source
And it's gonna be way cheaper than the upside down pyramid that the client actually uses.
Original source
All in all, this is a huge improvement, if it passes some pretty serious testing with various movement related spells (blink, charge etc).
rektbyfaith
09-23-2025, 03:11 PM #9Archived author: Pursche • Posted: 2025-09-23T15:11:22.895000+00:00
Original source
All in all, this is a huge improvement, if it passes some pretty serious testing with various movement related spells (blink, charge etc).