BUG cries on Bugfarm Fortress and Sumo

[Z-Man]

SuBlImInAl.FreQ pointed out that by doing the kind of trick that killed his teammates, he gains more speed. Duh, well, if you think that's worth it Anyway, the fake lag bug, at least as far as a regular client can trigger it, should be fixed now.

Code: Select all

[13] whatup: wow.
[5] Gusbo T.Jam esq --> Teammates: incoming
[7] psyko.Sonar: woah
[7] psyko.Sonar: nooo
[7] psyko.Sonar: WOAH
[7] psyko.Sonar: U FPAHSER
[2] ~*SP*~joda.bot: BUG
[0] whatup core dumped [spider]gizmo for 2 points.
[7] psyko.Sonar: lol
[0] ]i[ ~*King*~ core dumped Cycle_ for 2 points.
[3] [spider]gizmo: wtf
[7] psyko.Sonar: awesome
[3] [spider]gizmo: laggy as hell
[7] psyko.Sonar: look at that
[2] ~*SP*~joda.bot: BUG whatup went straight through the wall

Yes, this is a tricky one. whatup and gizmo were going parallel, whatup turned towards gizmo, and gizmo tunneled to the other side of whatup's wall. Otherwise, he'd have died immediately. This is beyond the limit of what the current engine is capable of handling, but there is one hack I can try after 0.2.8.2 is out.

Code: Select all

[5] Fonkay: BUG? I died for nothing

See attached plot. Maybe your command packet that turned you away from the wall was lost. The received command packets are indicated by the crosses.

[joda.bot]

@z-man: I thought of some way to get a hand on the imposter, but fail to remember it now ... or it was something else... I'll remember it ... and then use a private message for it next time

[Z-Man]

With the subversion migration coming up, I may take the liberty not to investigate the bug cries in the next days, probably until the servers get restarted as "SVN Test"

[Z-Man]

Code: Select all

[6] -CTP-$teffen voted against Poll "Kick V1adTHEImposter".
[13] [fe]belenus: BUG: Z-Man please do something against this imposter polling crap

I got to credit V1ad here for a creative naming choice. Of course, the corresponding IP is banned, and the ban script output is no longer put into everytime.cfg, but read by the server from standard input. That means it will be effective immediately.

That doesn't help against the drive by kickvoting V1ad was practicing here. I'll just add a time limit to votes. If you're not online for at least some minutes, you can't issue kick votes. Probably a couple of more things. Made a database entry for it.

Unfortunately, my attempts to get SVN working on the machine that runs my servers have been failures. Without convenient way to get the bugfixes back into SVN, the servers are useless. Maybe I'll just run Sumo from home then and leave Fortress finally to the real servers.

[Z-Man]

Code: Select all

[1] Lucifer: BUG
[1] Lucifer: spongebob just went through my wall

Turns out that if you aim really well or are really lucky, players pass through each other in direct cycle to cycle collisions. For the record, Lucifer should have died here. This is an old bug and probably won't be fixed for 0.2.8.2.

I've got SVN running here. I'll leave the spoon server and Fortress on CVS (because they're tuned to 0.2.8.2) for now, but switch Sumo over to SVN as soon as I can.

[Z-Man]

Code: Select all

[1] featherfcuk: wow..  i'm fastest?
[1] featherfcuk: i haven't even played yet
[1] featherfcuk: is that a BUG ?

A clientside bug, maybe. Tank, Wrtl, Lucifer? Probably need a clientside recording.

Code: Select all

[0] ~*SP*~joda.bot core dumped teammate bashusr! Boo! No points for that!
[5] ~*SP*~joda.bot: bash: did you turn left or right ?
[12] bashusr: left
[12] bashusr: i just turned too early
[12] bashusr: nps
[5] ~*SP*~joda.bot: bash: because u are left
[12] bashusr: yeah
[12] bashusr: i know
[5] ~*SP*~joda.bot: BUG bash might have died of teleport ?

No, just an enemy influence artefact. bashusr really crashed into some other wall (gableska's, I think), but the server assumed you pushed him to do it.

Vangoe is totally banned now. He did this before (joining with multiple clients in chatbot mode) and got forced his name to that as a result and warning.

Side note: Sumo is now running from SVN.

[wrtlprnft]

The fastest counter "bug" only exists in the branch. The counter gets reset with every new round, so he probably was fastest on some other server, left before the round ended and re- entered on CVS Fortress, and the fastest player there was slower than he was.
Same applies for the upper limit of the speed gauge, just it doesn't ever get reset, so if you play on a highspeed server, leave and join a lowerspeed server the gauge gets practically useless.

[Z-Man]

So it's fixed in the trunk? Could we get the fix backported or does it rely too heavily on the new stuff there?

[Lucifer]

If I recall the old code correctly, it would be something of a nasty hack. I seem to recall it being a static variable inside the function that computes and renders it. I suppose we could make it a global variable and just reset it to 0 in sg_HostGame and the other one.

[Lucifer]

I think I fixed it, maybe not. It passed my test anyway. I went to swampland and got the fastest player there, then went to local game, and when gameplay started it got reset.

If this doesn't totally fix it, the variable to grep for is max_player_speed. It's a global variable and max in gHud gets set to it.

[wrtlprnft]

So it's fixed in the trunk? Could we get the fix backported or does it rely too heavily on the new stuff there?

Actually it's not really fixed in the trunk since the code there displays the fastest speed any cycle on the grid has. So if the fastest player has a speed of 100m/s and drops down to 50m/s while someone else has a constant speed of 70m/s in the end the fastest speed is 70 m/s.

I't say port lucifer's branch fastest code back to the trunk and make a different callback out of it so the cockpit file author can choose which speed he wants to be displayed

[wrtlprnft]

Done "porting" (didn't do more than finding out the locations in gGame.cpp in the branch), there's two sets of callbacks now. The default file now uses the fastest per round meter, mine uses the fastest total meter since I find it more useful.

Also fixed the colors of the brake gauge

[Z-Man]

Code: Select all

[0] Sasha core dumped niro for 2 points.
[0] niro core dumped Sasha for 2 points.
[15] Sasha: LoL
[0] Sasha core dumped bashusr for 2 points.
[0] Sasha core dumped Vanhayes for 2 points.
[3] Vanhayes: ?
[3] Vanhayes: wtf just happened?
[3] Vanhayes: nothing was there
[3] Vanhayes: BUG died for no apparent reason

Sasha had adjusted to the rear rim wall. His death was 8.16 seconds before yours. Walls start to disappear 8 seconds after death and become harmless .2 seconds later. So it was a close call.

[Z-Man]

Code: Select all

[15] MaZuffeR: BUG shouldn't there have been a hole where i died?

Yes, you were exactly at the crash position of Nir. However, for some reason, Nir's explosion wasn't triggered there, but four grid units away in Nir's projected path, not touching any wall. That was because Nir's rubber detection failed and he was actually moved beyond the collsion position at death time. I hope the fix for that doesn't cause side effects.

Both test servers are now running from SVN sources and have been temporarily renamed to "Bugfarm".

[Z-Man]

Code: Select all

[0] Ocram core dumped \x for 2 points.
[0] KamP committed suicide.
[0] Match Winner: Team blue
[0] Overall Winner: Team blue with 102 points.
[0] \x core dumped Emlock for 2 points.
...
[7] \x: tied
[1] KamP: tie?
[11] Gusbo T.Jam esq: no way, it's a tie!!
[4] yea: NOPE
[6] alf ñ XP: 102-102
[4] yea: tie
[1] KamP: BUG we got a 102-102 tie and it chose a winner
[11] Gusbo T.Jam esq: no way, it's a tie!!

No, Team blue reached 102 points earler. To break ties, the scores are evaluated continuosly, not only after rounds.

We could add an option that the best team needs a certain point distance to the next team before it is declared match winner.