Gaming Zone > Unreal Tournament 2004

Freon bugs and issues to look at....

<< < (5/26) > >>

Piglet:
It wasn't sarcasm. You misunderstood.

The current equaliser is exactly as I specified. It moves the latest joining player from the team with 2 more players to the team with 2 fewer.

There is an option which we are not using 'Auto Balance When New Players Join' which would re-do the teams every time a player joins (and potentially when one leaves). This appears to be what you are suggesting.

Think what would happen... Each time someone joins the entire teams would be shuffled. You'd be moved from team to team ... Is that better than what we have at the moment? This isn't sarcasm, it's a question.

2. You have to remember that what you see on your screen is what your client game has calculated as having happened. The server takes all the input from each connected client machine and decides what actually happened. It can sometimes decide that what your client showed was wrong. This is why you see warping players as the server updates your client on the actual player position.

Where the  'New Net' (From utcomp I think) is switched on the calculations made to determine what happened are changed. I've not looked at the code, but I suspect that it biases the decision about what happened more towards what the high ping client machine calculated than using what the server calculated.

However the calculation is done there will always be times when the final decision about what happened isn't what you saw on your screen when taking information from any client at all. As soon as you give any credence to what happened on clients running 200+ later than the server then it's going to result in significantly noticeable updates.

Or as I put it before.. it's netcode problems. Unless you can get instantaneous communication between the client machines and the server then you have to predict on the client and correct all clients based on what the server decides.

hagis:
the 3spn balancer option isn't that bad,. it rarely moves the players who are doing well - which keeps things fairly consistent, balance on join doesn't need to happen - people can move themselves if they joined the wrong team (so long as the teams are uneven players) - when someone leaves and that team goes two players down I think the 3spn balancer should be run at that point - moving the last player can make things more uneven

I'm thinking of changing the logic to reward based on actual thaw work done.  - SoL did a version like that where they took into account thaw times,. it's on github but I didn't like the way they did it (and it broke the dual thaw distance code)

RFF - ok sure it's annoying when people jump in front but I'd count that as gameplay,. not pointing fingers but on other servers I see ppl complain about this and often they stand behind and in crowds of players shooting crap - so it's partly on them perhaps(!)

on netcode,. it's a fair chunk of work but when void added netcode to 3psn base code he pulling it from utcomp - lock stock and barrel,.. which is fine but the guy who wrote utcomp came back later and said hey I made tons of improvements to netcode in utcomp,.. the 3psn version of netcode is using the previous utcomp version,,

whether it's much better netcode is harder to prove but if you play on a never utcomp server across the pond it feels better than the netcode in 3spn - at least seems fairer

it's a distributed game and whet one player sees is not the same as another - it can't be perfect in a fps because there are network delays - unavoidable

hagis:
there is a bug that doesn't happen often but it's in all versions of 3psn,.. sometimes you get killed instantly when someone res's you (permadeath) - it seems to be linked to whether you're holding down the fire button when frozen at  the point you're res'd you die,.

bugs like that I'm not sure how easy to fix though(!)

btw - this happens on 3psn TAM servers as well so it's not freon specific but in the base code - and I wonder really if the cause is somewhere in the engine code,. but it probably could be coded round - like by adding a check if the player died within a second or two of being res'd (similar to how the timer works at the start of round to stop ppl from killing themselves at round start)

Piglet:
How much effort would it be to replace the netcode from the latest utcomp?

I've not seen the death on thaw

In the current version on the server I've added the option for counting 100 health units of thaw as a thaw, with a bigger reward every 50 and a slightly smaller one at each 100. It's not been tried out apart from on my test server - and it's not currently active on the Mia server. I might turn it on at some point over an evening to see how it goes. It counts all the partial thaws...

hagis:
yeah - it only happens on a res,. think if you try holding down the fire button you might see it - most ppl don't do that - but if you die and are res'd quickly it does happen

utcomp netcode change probably 5-10 hours,. it's just copying the right parts of the utcomp across / over the top of the existing,. 3210cw and onwards the netcode is sort of inside 3spn but in separate files so it's mostly just copying those over the top (and perhaps slight tweaks!)

Navigation

[0] Message Index

[#] Next page

[*] Previous page