Buggy settings...?

Self_Destructo · Post by **Self_Destructo** » Sat Jul 22, 2006 10:15 pm

No time to post or talk, so I just copied and pasted out of IRC...

<Self_Destructo> Ok, my server has been upgraded to 0.2.8.2 finally
<Self_Destructo> both .1 and .2 are very buggy....?
<Self_Destructo> I was wondering id it was possible for someone else to take the settings to my server and try to figure out if it is the settings that cause the wall bugs to happen
<Self_Destructo> is anyone here willing?
<Self_Destructo> no one?
<Self_Destructo> man, you guys probably hardly ever test your bug fixes and stuff with high speed and rubber settings do you guys?
<Self_Destructo> because the more fortress bugs you fix, the more high speed and rubber guys complain about bugs.
<Self_Destructo> ok, don't talk to me then

Self_Destructo · Post by **Self_Destructo** » Sat Jul 22, 2006 10:59 pm

Oh, I forgot to add what kind of bugs too... mostly wall bugs, where just just go up to a wall and die right off the bat even though you had a full rubber meter. I will try to get the guys to do some recording for us.

Post by **Lucifer** » Sat Jul 22, 2006 11:01 pm

I think the death without rubber usage is a bug currently being sought on the bugfarm servers. At least, I've seen commit messages that say "hopefully this fixes death without rubber".

If that's the biggest bug, we'll probably need server-side recordings to match the client-side recordings.

Website · Post by **Z-Man** » Sat Jul 22, 2006 11:16 pm

Yeah, recordings would help, don't forget to state the exact version they're recorded with.

The rubberless deaths should only happen in wild fights, when close to other players. They're fundamentally unfixable in the current architecture, I tink in current SVN I've done everything that's possible to make them as improbable as possible (pending further bugs in the implementation). One thing that can be done still would be to disable ping charity on the server. The clients' cooperation is required, unfortunately.

Self_Destructo · Post by **Self_Destructo** » Sat Jul 22, 2006 11:35 pm

Lucifer wrote:I think the death without rubber usage is a bug currently being sought on the bugfarm servers. At least, I've seen commit messages that say "hopefully this fixes death without rubber".

Ok, good to know.

Lucifer wrote:If that's the biggest bug, we'll probably need server-side recordings to match the client-side recordings.

It's currently not possible to get server side recordings, sorry.

Website · Post by **Z-Man** » Sun Jul 23, 2006 8:07 am

Oh, I forgot: rubberless death occur when cycle A hits the wall of cycle B, but it turns out that in game time, cycle B should have been there later, so B should have been the one getting stopped by rubber. If B did a turn in the meantime, that cannot be arranged; right now, B is killed instantly.
There would be two other options: We could let A pass through B's wall, or we could ignore the fact that A should have been there first legally and stop A. The latter may be the favorable option on high rubber servers; with low rubber, by the time the update about hitting the Wall arrives at A's client, the cycle is already dead, but with high rubber, he's got a chance. We could arrange that with an option.

Self_Destructo · Post by **Self_Destructo** » Mon Jul 24, 2006 11:44 am

I'm guessing lag is what causes it to get so far off in the first place?

Website · Post by **Z-Man** » Mon Jul 24, 2006 4:05 pm

Sort of, depends on your definition of Lag

Yes if it is "network latency", no if lag is "unusual peak network latency" for you. The different clocks of the individual cycles compensate the different regular network round trip times, AKA pings.

Self_Destructo · Post by **Self_Destructo** » Mon Jul 24, 2006 10:52 pm

Yeah, I meant network latency by the word lag. But now I'm very curious to know how you fixed this seeing that network latency will always be a problem. And I do think it would be wise to make it so we can chose how our server reacts to this problem.

Website · Post by **Z-Man** » Tue Jul 25, 2006 8:07 am

Yeah, read the "two other options" post two ago

We could make them available via setting variables.