Loading...
  OR  Zero-K Name:    Password:   

Bulk bug report

42 posts, 1108 views
Post comment
Filter:    Player:  
Page of 3 (42 records)
sort
- With either 253 or 254 players UI will not load. Tested with CAI. MYrankxponen and I were testing this earlier. This has something to do with Deluxe player list. Also ınterestıngly on balance this would cause massive lag.

- Odd bug with vindicator dgun. If you pick up something and just as it is being picked up and it dies, the vindicator will no longer be able to dgun units and will spam errors when you try to dgun again.

- Very rare bug with UI: seems on occasion bad commands will be added. Example: Unknown what causes this.

- Puppies on fight command with no targets in LOS will attempt to jink anyways. (Unconditional jinking) [Unknown if this is intended] - Intended behaviour

- Occasionally when building a mex thru the build command (not ctrl-w) the BA-like metal map will be stuck. - Seems fixed.

- Geothermal in pregame building has a broken texture. - fixed

- Spectators may occasionally hold up games while loading or prevent games from ending when joining. This seems to happen on occasion as well as in 200+ player games.

- Exit requires votes from players who aren't even involved or cannot vote.

Pet peeves:
- Horizontal Scroll bar in chat. Yuck. (got tired of this finally) - fixed.
+4 / -0


10 years ago
quote:
- Puppies on fight command with no targets in LOS will attempt to jink anyways. (Unconditional jinking) [Unknown if this is intended]

Also affects ticks and roaches, presumably all suicide units?
+1 / -0
quote:
Occasionally when building a mex thru the build command (not ctrl-w) the BA-like metal map will be stuck.

Is this during pre-game (initial queueing) or in actual game? because both ingame and pre-game have different widget that check for whether a player is building mex or not.

quote:
- Geothermal in pregame building has a broken texture.

Happy news. This was fixed (in test version) :)
+1 / -0
10 years ago
The bug with metal map happens for me every time i build a mex ingame. It's really annoying since i have to press F1 or sth twice after building every mex to fix it.
+1 / -0
10 years ago
Okay, this metalmap issue seem fixed in latest version. Happy news!
+1 / -0


10 years ago
quote:
Is this during pre-game (initial queueing) or in actual game? because both ingame and pre-game have different widget that check for whether a player is building mex or not.


Both. I believe.
+0 / -0

10 years ago
quote:
With either 253 or 254 players UI will not load

and i thought 10v10 was a clusterfuck
+5 / -0


10 years ago
quote:
With either 253 or 254 players UI will not load. Tested with CAI. MYrankxponen and I were testing this earlier. This has something to do with Deluxe player list. Also ınterestıngly on balance this would cause massive lag.
What is this? A number of players? A version number?

quote:
- Puppies on fight command with no targets in LOS will attempt to jink anyways. (Unconditional jinking) [Unknown if this is intended]
This is intended because they have such short sight range that they would not see the thing that killed them. Also when they fire they die so there is no normal fight command for them. This occurs for Roach and Tick as well. For them Fight is effectively the Jink command.

quote:
Pet peeves:
- Horizontal Scroll bar in chat. Yuck. (got tired of this finally)
I thought this was fixed. The simple fix for it is to revert everything done to the chat in the last week or so.
+0 / -0
^.^ ZK balancer is not optimised to balance games of 254 players

I demand my paired balance algorithm be implemented, cause it can handle 254 players much more easily (and its capable of doing case by case filtering of pairs it does not need to calculate (when elo1~=elo2, clans, ect))

:P curent balancer needs to go through 10^610 combinations
(lol so silly, the universe itself is only 10^82 atoms)


paired balancer would only need 10^269 (and produce a beter balance)
(asuming player elo is normal distribtuion, it could do it in as little as a million calculations due to filtering posibilities)
its 10^341 times better
+1 / -0


10 years ago
254 players is enough that you could just !random and expect it to be balanced.
+6 / -0
:P but the same rules would apply to 12v12 games, paired balancer would be (much) faster even without filtering (with filtering it would be much much much... faster)

actualy you could probably balance a 254 player game using a sequential paired method (what we do when we look at DNA). and it would only cost around 1000-2000 calculations in total, although admitedly less acurate.
+0 / -0

10 years ago
"curent balancer needs to go through 10^610 combinations "

ETA for this? Year? century?
+0 / -0
if I might add:
transports able to pick up enemy units, then the unit stops firing whatsoever like as though it doesn't know it's on an enemy transport.
this should either -require- the unit be emped before picking up, (as then it's actually disabled, not just standing still) or it takes damage when doing so (enemy unit fighting back).

urchins can be built on land.
1: it makes a torpedo launcher untargettable by torpedo launchers. this doesn't even make sense as torpedos would need to be -in- water to fire.
2: gameplay wise it's broke as well, as this makes urchins completely invincible to all but maybe 2 sea units, including ones that are obviously -supposed- to be a hard counter to them (sniper submarines)

in short, it makes the urchins -completely- overpowered compared to the rest of the sea units available.

-edit: or an even simpler solution... torpedoes launch themselves into the air to hit it, but will very quickly fall back to the ground making there range onto ground -extremely- limited (the same amount it takes an urchins torpedo to fall)
+0 / -0
10 years ago
Yes, please fix the metalmapbug.
+0 / -0

10 years ago
@bananaman, this was supposed to be a thread about actual bug reports, not balance complaints.
+2 / -0
after reading:
http://code.google.com/p/zero-k/source/browse/trunk/other/Springie/Springie/autohost/AutoHost_commands.cs

i stand enlightened (and with headache)

from what i can understand the curent balancer:
picks the biggest elo player, dumps it on the first available team, then cycles through the teams till it has no players left
(its basicly just one step above !random)


my algorithm would be slower then this, and it will definetly give beter balance


sidenote: why didn't you just do a sort step at the start on ranker List?
(instead of sorting through all the rankers evry time you are looking for the highest elo player)

also isnt:
quote:
foreach (var c in candidates)
{
var newElo = ((teamUsers[minid].Sum(x => x.Elo) + Math.Round(c.Elo/10)*10))/(teamUsers.Count() + 1);
if (newElo > maxElo)
{
maxUsers.Clear();
maxUsers.Add(c);
maxElo = newElo;
}
else if (newElo == maxElo) maxUsers.Add(c);
}
var pickedUser = maxUsers[rand.Next(maxUsers.Count)];


doing the same thing as:
quote:
foreach (var c in candidates)
{
var newElo = Math.Round(c.Elo;/10)*10
if (newElo > maxElo)
{
maxUsers.Clear();
maxUsers.Add(c);
maxElo = newElo;
}
else if (newElo == maxElo) maxUsers.Add(c);
}
var pickedUser = maxUsers[rand.Next(maxUsers.Count)];

just with a tone of usless steps?
(you can also take out the randomiser and Math.round() steps because they also have no real effect on the balance)
+0 / -0
quote:
ZK balancer is not optimised to balance games of 254 players


I thought it had been improved and uses Chase's twiddle algo (or similar)... looking at zk's code makes me angry everytime... especially when there are multiple Balancer classes and we need to find out which one is used...

quote:
paired balancer would only need 10^269 (and produce a beter balance)


That is still pretty crappy, considering there are less than 10^77 valid combinations...



http://www.wolframalpha.com/
p=254,b=2,sum(binomial(p,k)),k=ceil(p/(1+b)) to (p/2)
+0 / -0
BRrank[V]sheep sheep see the post above you

i was simply using 254^254 vs 128^128 (really really crude math), ignoring how you would actualy code things

now that i looked at the code curent balancer does something like
7*(254+1)*128=228480 calculations +a few from other parts
(although it only really needs to do 32640, but code is weird)


for all practical purposes my metod would have to do at least 32640 calculations for sort, and 128 for balancing
(although realisticly it would probably do more along the lines of 1280 for balancing because i lika the accurate)
+0 / -0
10 years ago
sry, I'll make a new thread for those then... I posted it in here because to me, those ARE bugs... or at the very least exploits. and didn't want to create multiple threads when this was supposed to be a list of said bugs.

but anyway, if you want them in there own thread, fine. last thing I want is to be the cause of an argument in a bug thread :/
+0 / -0

10 years ago
USrankbanaman

You can be forgiven for not knowing the long and sorry history of comnapping, but the torpedo launcher actually changes models completely as it goes onto land. It's painfully obvious that this is intentional, and not a bug.

Intentional behavior means not bug. Full stop.
+0 / -0
Page of 3 (42 records)