16 replies · 688 views · thread synced · 4 days ago
· View on torn.com
About this thread
Posts archived:17 / 17 posts (100%) · the total is Torn's reply count + the opening post at the last fetch
Counted by TornLife from the archived posts.
Archived posts
17
Discussion span
→
People posting
6
Likes on archived posts
32
Posts by staff, officers and moderators
5
Authority score
61 / 100
Historical score
41 / 100
Story score
50 / 100
Engagement score
67 / 100
Most-liked replies
Aezur[1950769]
· 11 likes · This is a safety valve, not a feature from my understanding. Attacks put locks on the DB, and disallow some in-game actions. The timeout releases the locks, and allows the game to continue as normal should something unexpected happen. Using …
Chedburn[1]Admin
· 11 likes · A change has been made so that as long as you're waiting on the page, a poll request will be a made at exactly 0 seconds which will immediately 'Leave' the target then and there. For now though, if the …
Chedburn[1]Admin
· 2 likes · I'm afraid letting attacks timeout like this provides no guarantee of their precise ending time if you leave the page - it'll likely always be a few seconds afterwards. Even if you stay on the page, it could still be …
as you can see the fight ended at 00:39:04 and should of timed out at 00:39:34 which was 4min 59sec from the last attack but for some reason it did not auto leave at 30 and added 4 seconds causing the chain to end
No we did auto/timeout. After the last attack a timer starts for 30 seconds where it auto leaves. Weve been doing this for over a year and is how weve always stalled chains. The attack was already over and the chain clock still had 31 seconds so we left it to auto timeout which weve done hundreds of times in the past. Normally when it's this close the chain would fake end but if you refreshed the page or waited a second it would be just fine counting down for the next hit. This is the first time I've ever had a chain break doing this strat and weve regularly let them auto leave around 31-40 chain clock in the past without ever having an issue
for clarity, we use auto leave because as far as we were aware its a flat 30 second timer that the attack ends if no choice is made. doing this lets us safely stall a chain since if the player is attacked it just auto leaves early rather than them getting hit mid attack trying to make a hit in the last few seconds, it also lets everyone see in the faction attack feed that someone has already made the chain save hit while letting us know 30 seconds earlier so we dont have last second panics. this is the first time ive ever seen the timer actually take longer than 30 seconds (it is an automated process that requires no player input so its rather weird to be showing a counting timer of 30 seconds that end results in a 34 second automated process, all of which is shown in the time stamps)
I do autoleave too, going to single digits and I know torn is SUPPOSE to reset chains back to where they should've been if torn servers mess up, or other pieces of the game that'd break the chain. (Lag, delay, or in this case auto leave, or the chain timer being wrong.) That's why I feel secure in doing such high risk plays when it's next to none.
Now, I'm not sure if the persons internet, or whatever needs to ping the server that 30 seconds are up or not, that might've caused the delay.
But yeah, I'm curious on what this problem could be. This would be valuable to know. 31 seconds passed, not 30. (Unless it was 30 and a half, and they hit at a half/8th, whatever of a second. Causing the server to count that second as a extra.)
since it seems it may be a while since i hear back again ill add this question
in a situation like this what happens if this is deemed a bug this chain was days ago so whats the outcome? like do we get a chain restarted where this left off or are we just SOL. if this isnt deemed a bug whats the standard time variance of the auto timer than because if the timer exists with a countdown down of 30 seconds but doesnt match the visual then whats the limit at which this is considered reportable, for example if we continue to use this to stall from 41-50 seconds and it takes 50 seconds to time out is it still considered SOL
I don't think we have restored a chain like this. We do have another report on this here. The main issue here is that the 30s auto-leave sometimes fails, and I am not certain if it is fine for the factions to rely on the auto-leave to keep the chain running.
I will pass this along to Joe to check his thoughts on this.
I'm afraid letting attacks timeout like this provides no guarantee of their precise ending time if you leave the page - it'll likely always be a few seconds afterwards. Even if you stay on the page, it could still be a couple seconds afterwards. Technically it would be a challenge for us to synchronize these timeouts to happen on the exact second they're due - without using a purpose-built service to handle them, which maybe we will do one day.
would it be possible to at least get an out of variance limit so that oddities or clear backend issues could be fixed/chain restarted in the future?
like could we say if its 40 seconds or greater and the last hit was 40 seconds from chain end that this can be deemed an irregularity and fixed? i ask this partly do to the other ticket claiming to have clicked leave but finished from auto timeout instead, unless im mistaken i assume theres no proof on if someone clicked or did not click leave on fights that auto leave without the player screen recording (seeing as auto is the backends forced outcome if it doesn't receive the player outcome)
would be nice to have something deemed safe to lean on to handle external issues. as mentioned earlier this is how we see if someone has finished their attack and weve relied on this system since once it reaches "attacked" in the chain feed weve known the hit was made and whether was safe or not via the timer, having a baseline lets us know we are safe even if the player lost internet or couldnt click leave from actual issues otherwise wed all have to attacked every 5 min since there would be no assurance that an attack shown as complete would ever time out
Yeah, we relied on this function a lot during low population hours while chaining. Lots of e would be wasted if this isn't deemed reliable.
Telling folks not to rely on game functionality delivering as described is kind of a bad look, so it may warrant attention at some point, once higher priorities have been attended to.
It's very difficult to synchronize it perfectly, and obviously it's preferable to have it slightly longer if necessary rather than timing out before it says it will.
This is a safety valve, not a feature from my understanding. Attacks put locks on the DB, and disallow some in-game actions. The timeout releases the locks, and allows the game to continue as normal should something unexpected happen.
Using it to chain with seems... foolish? It's like putting your hand in a band-saw and complaining the safety catch didn't fire. Like, yeah, it should have fired, but also you shouldn't have shoved your hand in the saw.
Edit: I do understand the frustration. Just offering an alternative viewpoint.
Requirement- Make sure that the user has 30 full seconds after the finishing hit to select target disposition, then auto select Leave if nothing is received. This needs to work even if the user navigates away.
If the user navigates away, then there is no conflict on timestamps- it should just go when server time reaches 30 seconds.
This doesn't work for a situation where there is a disconnect/reconnect, so it would also need the use the following:
If the user stays on the page, the local timer starts after the server response from the finishing hit is received and is therefore behind the server. So the server side has to give a little extra to give the user 30 full seconds. Which results in a problem if the user has the chain timer burn down to a time inside that slush period.
Well, I see the issue. I may have a solution for you. If the user allows the autoselect to determine target disposition, then reset the chain counter based on the finishing hit timestamp or 5 seconds after the finishing hit timestamp. Then if people care about stretching it out, they will have to hit the button instead of leaving it on the system.
If we use your analogy, players are let into the shop with no prior experiemce in a wood shop and told to try things to figure out what works. So the player in question has probably seen dozens of times when putting their hand in the bandsaw was caught by a guard, which for some reason failed in this case.
The game includes very little formal direction to players, so it falls on the developers to anticipate what players will do or to be willing to live with the consequences when they try things. This may call for a little flourescent yellow tape as it were.
A change has been made so that as long as you're waiting on the page, a poll request will be a made at exactly 0 seconds which will immediately 'Leave' the target then and there. For now though, if the page is left, it may be a few seconds late.