DonKosak
Joined: Apr 06, 2004
|
 Posted:
Jun 02, 2005 - 11:07 |
 
|
| Tommi wrote: | | and for the rest of the population it doesnt matter because they cant do it. |
Hopefully it doesn't matter because everybody knows that it would destroy the Fumbbl-fun... |
|
|
Rynkky
Joined: Aug 03, 2004
|
| Posted:
Jun 02, 2005 - 11:21 |
|
| Tommi wrote: | | The random seed is global between VM:s? |
Seed is instance based, so it is not global.
| Tommi wrote: | | Or the same VM runs all different apps? |
AFAIK new VM is created and started for each java command. If application starts new applications, they will be run in same VM as the parent application (this is just guessing, might be very wrong!).
| Tommi wrote: | | Or the client does not reset the seed? |
Client resets the seed.
| Tommi wrote: | | How does it actually work? |
Maybe when these issues have been fixed I will publish the source and explain the technique behind. |
|
|
JanMattys
Joined: Feb 29, 2004
|
| Posted:
Jun 02, 2005 - 11:55 |
|
My head hurts.
 |
_________________
 |
|
DoubleSkulls
Joined: Oct 05, 2003
|
| Posted:
Jun 06, 2005 - 06:45 |
|
When you initialise the random number generator you can specify a seed value (by default it uses the current time in milliseconds since 1970). Once initialised with a given number any java random number generator will produce exactly the same sequence of random numbers (this is immensely useful in some circumstances where you need repeatable values).
Presumably the host determines a seed value and then gives the seed to the client. They then will produce exactly the same sequence of numbers - as would any other java program with the same seed. If you can read the message sent to the client and know the seed you can easily predict the next number is the sequence. Its probably the most obvious programming cheat on FUMBBL and TBH I'd be surprised if Rynkky was the first to do it.
Although Rynkky's honesty and willingness to help are admirable I suspect his reticence over the mechanics are a little unnecesarry. Any java programmer who had spent 10 seconds thinking out it would have realised this was the easiest way to cheat without detection. |
_________________
Ian 'Double Skulls' Williams
Eucalyptus Bowl |
|
Laviak
Joined: Jul 19, 2004
|
| Posted:
Jun 06, 2005 - 08:50 |
|
Its easy enough to prevent people intercepting the seed (or at least make it very difficult to do so). All that is really needed is to encrypt the seed when it is sent.
The problem is that this will only stop hacks that intercept the message containing the seed. Encrypting the message won't mean a thing if you are able to extract the seed directly from the Java VM, and I think that it is possible to do so (though I can't remember for certain). Can anyone with sufficient java experience confirm whether this is the case? |
_________________
We Fink Wer Orks
--------
Help save blood bowl, foul an elf today!. |
|
Rynkky
Joined: Aug 03, 2004
|
| Posted:
Jun 06, 2005 - 12:04 |
|
| DoubleSkulls wrote: | | Although Rynkky's honesty and willingness to help are admirable I suspect his reticence over the mechanics are a little unnecesarry. Any java programmer who had spent 10 seconds thinking out it would have realised this was the easiest way to cheat without detection. |
I will write a small essay of how it is done. But I won't do it until client has been fixed because I will also write about flaw in the client.
SkiJunkie has the demo program and source code + my description so I think we will get update soon (at least I hope so).
-Rynkky |
|
|
DM
Joined: Aug 13, 2004
|
| Posted:
Jun 11, 2005 - 04:20 |
|
Well after the last ranked game I've just played I have to admit I'm a true believer on some of the comments posted on how the shared seeding screws the game. The 'official' luck stats say that the game was actually in my favour, but if you watch the replay you'd easily scoff at those 'statistics'.
Infuriating when you're on the receiving end of it. |
|
|
BadMrMojo
Joined: Aug 02, 2003
|
| Posted:
Jun 11, 2005 - 05:36 |
|
Uhm... did you actually read any of the previous posts? The luck meter has nothing to do with the discussion and the shared key is responsible for a 2/10,000,000,000 discrepancy.
The potential problem being discussed about the shared key is something totally different from bad rolls.
Also, regarding your own game, a quick viewing of the replay shows that he did indeed get a few of lucky one-die blocks (the second TD, for example). I can't see why that's so surprising when 1/2 the block results will result in your thrower going down (the one without block, at least). The wardancers made 2 or 3 leap rolls and I didn't notice any that failed. That's about par for the course, slightly lucky (if he made 3). You were both failing lots of GFIs and I think your luck problems may have been exacerbated by the one or two turns which started with a 3+ dodge instead of standing up and moving free players first (in particular, the aforementioned 2nd TD).
On the other hand, you were making some fantastic pickup rolls with ag3 and your armor rolls were outright brutal - hence the full KO box. Looks like a moderately unfortunate match but hardly something noteworthy, in my humble opinion. |
_________________
Ta-Ouch! of BloodBowl
Condensed Guide for Newbies |
|
DM
Joined: Aug 13, 2004
|
| Posted:
Jun 11, 2005 - 07:39 |
|
You've been playing a lot longer than me mate, so it may not surprise you... but it sure frustrated the heck outa me. Some of the dodge rolling had my jaw on the flaw. I dunno... seemed a little 'arrrrrgh' from my perspective. *shrugs* Maybe just an unlucky day? Maybe I'm used to having more luck and this stood out? Dunno... |
|
|
|
|
| |
Jump up on a tree?
Search
