0001025: Doors don't respond after pressing +use online

Notes
(0004676) Torr Samaho (administrator) 2012-09-15 19:51	I have observed this problem too (and I suspect it started with some ZDoom updates I made long ago, i.e. before 98a), but without a way to reproduce the problem I can't do anything about it.

(0004680) ZzZombo (reporter) 2012-09-16 02:33	What I usually do when I experience the bug is just pressing +USE on a door online. Maybe you need a certain ping (0000064:0000200-300)? But even in such a case sometimes doors open normally, but with a significant delay.

(0004691) unknownna (updater) 2012-09-16 10:54	Quote from Torr Samaho but without a way to reproduce the problem I can't do anything about it. Could it be related to the +attack desync where the clients, very often, don't know whether you've pressed +attack? I don't need a ping emulator to reproduce the +attack desync.

(0004692) Torr Samaho (administrator) 2012-09-16 11:10	If the +attack desync only affects the other clients, then this is most likely not related. Can you post a link to the corresponding tracker ticket?

(0004700) unknownna (updater) 2012-09-16 12:15	Quote from Torr Samaho Can you post a link to the corresponding tracker ticket? It seems that the issue never had its own ticket. Here's the new ticket.

(0004737) Watermelon (developer) 2012-09-21 14:41	This would be a critical bug to fix as it's cost me duel games when the door doesn't open, then I die and the game dynamic changes after that. I always wondered how this happens. Is it because of a dropped packet? (possibly use RUDP if this is the case?)... or is it faulty interpolation on the server (thinking you're not where you should be and thus triggering nothing)...etc

(0004771) Torr Samaho (administrator) 2012-09-23 18:54	I had an idea what could be causing this: If you only hit +use very briefly, so that the client only sends a single packet with "+use" pressed and then either this packet gets lost or the next packet where "+use" is not pressed anymore arrives at the server during the same tic as the one with "+use" (essentially override the "+use" packet) this could cause the issue. Do doors open reliably if you keep "+use" pressed for a little while (like 0.5 seconds)?

(0004791) unknownna (updater) 2012-09-24 06:52	Quote from Torr Samaho Do doors open reliably if you keep "+use" pressed for a little while (like 0.5 seconds)? No, it seems that they don't. I noticed this in MAP12 and MAP15. The door at the beginning in MAP15 would never respond. IIRC, I had to press +use 5 times. I was playing with 3-4 other players on a survival server. My ping was fluctuating between 220-250. One thing I noticed in MAP19 is that whenever a +use attempt fails, the switch-doors will only play the first switch sound, but not display the texture animation or open the door and play the second switch sound. I must say that it happened very frequently. But I can't seem to reproduce the issues on my own server with an emulated ping of 220-250. But I managed to reproduce the issue on another server and recorded a demo of the switch-door issue. Please have a look at it.

(0004794) Torr Samaho (administrator) 2012-09-24 19:13	Quote from unknownna One thing I noticed in MAP19 is that whenever a +use attempt fails, the switch-doors will only play the first switch sound, but not display the texture animation or open the door and play the second switch sound. This is an interesting observation. I will look into this in detail, most likely on the weekend.

(0004835) Torr Samaho (administrator) 2012-09-28 19:34	The demo sheds some light on what's going on and have an idea how to fix this (if my current theory on what's going on is correct).

(0004843) Torr Samaho (administrator) 2012-09-29 17:10	Please test if this fixes the door problem without breaking anything else.

(0004858) unknownna (updater) 2012-09-30 07:19	Quote from Torr Samaho Please test if this fixes the door problem without breaking anything else. Could you tell us what you fixed in that binary? I did some tests in Doom2, AOW2 and WhoDunIt with the binary and it seemed to work fine. I can't say for sure if it fixed the issue. As I said earlier, I pretty much cannot reproduce the bug on my own servers.

(0004860) Torr Samaho (administrator) 2012-09-30 07:38	Quote from unknownna Could you tell us what you fixed in that binary? Sure. The problem is not that the server doesn't recognize +use, I could clearly see this from the demo: Whenever you pressed +use, the server told the client to start opening the door, but immediately after this it told the client to stop the door movement. This is what I know for sure from the demo and my theory why this happens is as follows: When the server gets two client movement commands where +use is pressed in one tic, it starts to open the door when processing the first command and closes the door when processing the second command assuming the client pressed +use again because "oldbuttons" is not updated properly after the first command is processed. What I did to fix this is to update "oldbuttons" after processing a client movement command. This way the server sees that the button was pressed before and doesn't close the door when the second command arrives.

(0004877) unknownna (updater) 2012-09-30 11:38	Quote from Torr Samaho What I did to fix this is to update "oldbuttons" after processing a client movement command. This way the server sees that the button was pressed before and doesn't close the door when the second command arrives. How many client commands can be squeezed into a single tic? What if the server receives more than 2 client commands where +use is pressed in one tic? Will this fix also handle that?

(0004900) Torr Samaho (administrator) 2012-10-01 20:07	Quote from unknownna How many client commands can be squeezed into a single tic? In theory there is almost no limit, but in practice I'd say that it's unlikely that more than two arrive during one tic. Quote from unknownna What if the server receives more than 2 client commands where +use is pressed in one tic? Will this fix also handle that? The fix has no problems handling more than two tics.

(0005188) Watermelon (developer) 2012-10-24 13:06	I added this to my own build that I use online with people, and I can confirm that it works really well. I haven't seen a single door jam yet or heard about it from anyone. I had euro's playing on the server and they had no problem with their 90-100+ ping.

(0005197) unknownna (updater) 2012-10-25 01:32	Torr, shouldn't we also be able to confirm that the fix works by re-backporting the player_t::usedown fix?

(0005807) Dusk (developer) 2013-01-21 16:20	'http://zandronum.com/forum/showthread.php?tid=2001#pid24542 [^]'

This issue is already marked as resolved. If you feel that is not the case, please reopen it and explain why.
Supporters:	unknownna ZzZombo
Opponents:	No one explicitly opposes this issue yet.

Issue History
Date Modified	Username	Field	Change
2012-09-10 11:57	unknownna	New Issue
2012-09-10 11:57	unknownna	Status	new => confirmed
2012-09-15 19:51	Torr Samaho	Note Added: 0004676
2012-09-16 02:33	ZzZombo	Note Added: 0004680
2012-09-16 10:54	unknownna	Note Added: 0004691
2012-09-16 10:55	unknownna	File Added: screenshot.png
2012-09-16 11:10	Torr Samaho	Note Added: 0004692
2012-09-16 12:12	unknownna	File Deleted: screenshot.png
2012-09-16 12:15	unknownna	Note Added: 0004700
2012-09-21 14:41	Watermelon	Note Added: 0004737
2012-09-23 18:54	Torr Samaho	Note Added: 0004771
2012-09-24 06:51	unknownna	File Added: 2012.09.24_08.47.13_doom2.cld
2012-09-24 06:52	unknownna	Note Added: 0004791
2012-09-24 19:13	Torr Samaho	Note Added: 0004794
2012-09-28 19:27	Torr Samaho	Assigned To	=> Torr Samaho
2012-09-28 19:27	Torr Samaho	Status	confirmed => assigned
2012-09-28 19:34	Torr Samaho	Note Added: 0004835
2012-09-29 17:10	Torr Samaho	Note Added: 0004843
2012-09-29 17:11	Torr Samaho	Status	assigned => needs testing
2012-09-30 07:19	unknownna	Note Added: 0004858
2012-09-30 07:38	Torr Samaho	Note Added: 0004860
2012-09-30 11:38	unknownna	Note Added: 0004877
2012-10-01 20:07	Torr Samaho	Note Added: 0004900
2012-10-24 13:06	Watermelon	Note Added: 0005188
2012-10-25 01:32	unknownna	Note Added: 0005197
2013-01-09 23:20	Dusk	Target Version	=> 1.1
2013-01-21 16:20	Dusk	Relationship added	related to 0000098
2013-01-21 16:20	Dusk	Note Added: 0005807
2013-01-21 16:20	Dusk	Status	needs testing => resolved
2013-01-21 16:20	Dusk	Fixed in Version	=> 1.1
2013-01-21 16:20	Dusk	Resolution	open => fixed
2018-09-30 20:53	Blzut3	Status	resolved => closed