MantisBT - Zandronum |
| View Issue Details |
|
| ID | Project | Category | View Status | Date Submitted | Last Update |
| 0002294 | Zandronum | [All Projects] Bug | public | 2015-06-08 20:46 | 2025-02-16 11:42 |
|
| Reporter | unknownna | |
| Assigned To | | |
| Priority | high | Severity | major | Reproducibility | always |
| Status | feedback | Resolution | reopened | |
| Platform | | OS | | OS Version | |
| Product Version | 2.0 | |
| Target Version | | Fixed in Version | 3.1 | |
|
| Summary | 0002294: Very evident misprediction of movement on raising floors in general |
| Description | I've noticed this issue for a while now and it seems that the client greatly mispredicts its movement if stepping up to a raising floor online.
In particular, this misprediction seems to occur when the player moves from a lower floor to a raising floor that is higher.
It seems to correct itself if you press +jump. |
| Steps To Reproduce | 1. zandronum -file stairs_02.wad -host
2. Connect a client to the server with an emulated ping of 200.
3. Join the game.
4. Move to the floor in front of you and move in a circular pattern with +forward, +back, +moveleft and +moveright. |
| Additional Information | It's very evident even with only 145 ping, which IMO is a relatively normal ping. And I always notice this issue when playing online on regular servers.
It started to break in 1.1-alpha-r121014-1138M.
MOVINGFLOORSCREENSHAKE.wad |
| Tags | No tags attached. |
| Relationships | | related to | 0002035 | closed | | Less evident player desync on a gap between raising floors | | related to | 0001682 | confirmed | | Floor_Waggle produces jittery lag online. | | related to | 0002293 | closed | Edward-san | Clients build stairs without any delay between steps causing unnecessary prediction issues online | | related to | 0001079 | closed | Torr Samaho | Client greatly mispredicts its z position if pushed over a ledge by something else | | related to | 0003334 | assigned | Kaminsky | Tickrate discrepancies between clients/servers | | has duplicate | 0002165 | closed | | Screen shaking w/ moving floors online | | related to | 0002401 | confirmed | | Floor_Waggle doesn't reset floor to its proper position online | | related to | 0001131 | feedback | | Client mispredicts its movement if moving on floors/slopes with negative height values |
|
| Attached Files |  stairs_02.wad (3,168) 2015-06-08 20:46
https://zandronum.com/tracker/file_download.php?file_id=1535&type=bug |
|
| Issue History |
| Date Modified | Username | Field | Change |
| 2015-06-08 20:46 | unknownna | New Issue | |
| 2015-06-08 20:46 | unknownna | File Added: stairs_02.wad | |
| 2015-06-08 20:46 | unknownna | Status | new => confirmed |
| 2015-06-08 20:47 | unknownna | Relationship added | related to 0002035 |
| 2015-06-10 12:01 | Frits | Note Added: 0012631 | |
| 2015-06-10 12:42 | unknownna | Note Added: 0012633 | |
| 2015-06-10 12:45 | unknownna | Note Edited: 0012633 | bug_revision_view_page.php?bugnote_id=12633#r7355 |
| 2015-06-10 12:45 | unknownna | Relationship added | has duplicate 0002165 |
| 2015-06-10 13:08 | unknownna | Note Edited: 0012633 | bug_revision_view_page.php?bugnote_id=12633#r7356 |
| 2015-06-10 13:09 | unknownna | Note Edited: 0012633 | bug_revision_view_page.php?bugnote_id=12633#r7357 |
| 2015-06-10 14:18 | Leonard | Note Added: 0012636 | |
| 2015-06-10 19:47 | unknownna | Note Added: 0012641 | |
| 2015-06-10 19:47 | unknownna | Relationship added | related to 0001682 |
| 2015-06-10 19:48 | unknownna | Relationship added | related to 0002293 |
| 2015-06-10 21:32 | unknownna | Note Edited: 0012641 | bug_revision_view_page.php?bugnote_id=12641#r7368 |
| 2015-06-12 21:23 | unknownna | Note Added: 0012666 | |
| 2015-07-10 22:11 | unknownna | Note Added: 0012851 | |
| 2015-07-10 22:11 | unknownna | Relationship added | related to 0001079 |
| 2015-07-10 22:14 | unknownna | Additional Information Updated | bug_revision_view_page.php?rev_id=7574#r7574 |
| 2015-07-10 22:25 | unknownna | Additional Information Updated | bug_revision_view_page.php?rev_id=7575#r7575 |
| 2015-08-12 15:46 | unknownna | Relationship added | related to 0002401 |
| 2018-08-23 05:09 | unknownna | Relationship added | related to 0001131 |
| 2024-03-07 10:57 | unknownna | Note Added: 0023309 | |
| 2024-03-07 10:57 | unknownna | Status | confirmed => closed |
| 2024-03-07 10:57 | unknownna | Resolution | open => fixed |
| 2024-03-07 10:57 | unknownna | Fixed in Version | => 3.1 |
| 2024-06-25 23:00 | unknownna | Note Added: 0023768 | |
| 2024-06-25 23:00 | unknownna | Status | closed => feedback |
| 2024-06-25 23:00 | unknownna | Resolution | fixed => reopened |
| 2024-06-25 23:01 | unknownna | Note Edited: 0023768 | bug_revision_view_page.php?bugnote_id=23768#r14286 |
| 2025-02-15 23:07 | unknownna | Note Added: 0024213 | |
| 2025-02-15 23:07 | unknownna | Status | feedback => new |
| 2025-02-15 23:07 | unknownna | Relationship added | related to 0003334 |
| 2025-02-15 23:07 | unknownna | Status | new => feedback |
| 2025-02-16 11:42 | unknownna | Note Edited: 0024213 | bug_revision_view_page.php?bugnote_id=24213#r14464 |
|
Notes |
|
|
(0012631)
|
|
Frits
|
|
2015-06-10 12:01
|
|
|
|
|
(0012633)
|
|
unknownna
|
2015-06-10 12:42
(edited on: 2015-06-10 13:09) |
|
Yeah, it's pretty much the same issue. And no, it's not quite the same for lowering floors (though it'll still mispredict if you jump). The jitter is very evident with rising floors in general.
It doesn't jitter on rising floors if you stand on them when they start to move (though you can still make it mispredict by running on the ledges or by jumping), but if you move to them from a floor that is lower in height, the jitter will appear.
|
|
|
|
|
|
|
|
(0012641)
|
|
unknownna
|
2015-06-10 19:47
(edited on: 2015-06-10 21:32) |
|
I'd say that they're kind of related without being the same issue.
Quote from Torr Samaho
While predicting, the client assumes that sector heights do not change during the prediction window. There are some additions that take care of moving floors, but these workarounds cannot cover the case where a moving floor changes its direction during the prediction window.
The difference between Floor_Waggle and this issue is that the rising floors never change direction here.
|
|
|
|
|
|
This worked fine in 1.0 and started to break in 1.1 so it's another regression. |
|
|
|
|
|
|
|
|
I can't reproduce this anymore in 3.1. The client quickly corrects itself now. It muse have been fixed by something else.
Thanks! |
|
|
|
(0023768)
|
|
unknownna
|
2024-06-25 23:00
(edited on: 2024-06-25 23:01) |
|
After testing some other wads, I noticed this issue still happens.
But there's a difference in 3.1 and beyond: If you get a certain tick sync it doesn't glitch up. And strangely enough it doesn't glitch when you get what I would consider a "bad" tick sync with lots of jitter. Getting a clean tick sync causes it to break.
And another observation is that playing with cl_capfps 0 might cause the client to correct itself, whereas with capped FPS it breaks completely until the floor has risen fully.
|
|
|
|
(0024213)
|
|
unknownna
|
2025-02-15 23:07
(edited on: 2025-02-16 11:42) |
|
I gave this another test using the newer clock sync build, and can confirm that the issue is caused by the tickrate desync between the client and server, in particular the tick offset the client's time lands on.
Assuming that you connect with a ping of 160 with stairs_02.wad,
using the movebuffer with "cl_usemovebuffer 1" masks the underlying issue, but when you use "cl_usemovebuffer 0" to check how good the tick sync really is, you always have to offset the client's time manually up by approx. 4-6 milliseconds to make the sync work right and prevent the desync from occurring.
This suggests that the current clock sync algorithm either isn't aggressive enough, or that it currently aims to land at the wrong spot, i.e., it currently aims to land at the middle of a server tick instead of the beginning or something.
What do you think?
Edit:
Or maybe the client prediction is simply making a wrong assumption somewhere.
|
|