User Pass
Home Sign Up Contact Log In
Forum > Discuss GLB Issues With Catch22 > NOT A BUG - CLOSED > Calling play that is NOT in my playbook.
Page:
 
geekor
offline
Link
 
The only thing I can think is it reverted to basic settings, but seriously why? I will have the owner give a bug mod access to double check if need be. I will put screen shots as well, as I change up my plays every game anyways. This was a scrimmage, but still a bug.

http://goallineblitz.com/game/replay.pl?game_id=695790&pbp_id=4175356

I (and who would) never play 3-4 vs 3 WR sets. I have my AI set by formation and down and distance. No single back or 3WR.

Screen shot of Def AI tester:
http://img38.imageshack.us/img38/2217/testerx.jpg

Of the 2 possible outputs in the DPC, Cover HB Dime:
http://img190.imageshack.us/img190/7552/coverhbdime.jpg

and 335 Base:
http://img196.imageshack.us/img196/391/335base.jpg

No 3-4.

This isn't the first time this has happened, a play being called that shouldn't.

Anyone bug mod want to go in and check, I will leave everything as is for today.
 
tragula
title
offline
Link
 
Do you trust the "clock running" ? It usually work, but maybe there are bug there. Like after a sack the clock stops.
(I have hunch that your input doesn't require the clock to run).
 
Captain Kaos
offline
Link
 
Do you have auto-adjust turned off? Follow AI exactly?

Any 3-4 play in the D AI at all?
 
InRomoWeTrust
Lead Mod
offline
Link
 
geekor while this is semi-irrelevant, I'd suggest adding a "catch all" at the base of you DAI as a safeguard against these type of slip through the crack type plays. I know I had a similar result, but my "catch all" caught the play that for w/e reason didn't match against any of my input/outputs. My catch all has only been called once from what I know, but it's saved me.

A conservative 3-3-5 C0 is a good idea for a catch all just to cover your butt for all situations. The last thing you want is an open WR.
Edited by Mat McBriar on Aug 19, 2009 16:15:16
 
geekor
offline
Link
 
no noob stuff like auto adjust, and clock running isn't used.

I've caught ones like this once or twice a season, and I get crap from the mods like are you 100% sure another GM didn't touch it, and other excuses. I've checked and doubled check everything, tested, screen shot everything.

A catch all might help, something at the very end just in case, but it shouldn't come to that. The simple fact is, the AI should call something and it didn't, I have the screen shots there, and will gladly allow a bug mod to come in and check. the fact is yes it happens less than 1% of the times, but it should be ZERO. It isn't and it's faulty programming, and can cost people games.
 
Captain Kaos
offline
Link
 
Originally posted by geekor
no noob stuff like auto adjust, and clock running isn't used.

I've caught ones like this once or twice a season, and I get crap from the mods like are you 100% sure another GM didn't touch it, and other excuses. I've checked and doubled check everything, tested, screen shot everything.

A catch all might help, something at the very end just in case, but it shouldn't come to that. The simple fact is, the AI should call something and it didn't, I have the screen shots there, and will gladly allow a bug mod to come in and check. the fact is yes it happens less than 1% of the times, but it should be ZERO. It isn't and it's faulty programming, and can cost people games.


Mods have to ask that because it's general questioning before verifying something is a bug. Not everyone thinks about all the different situations, therefore they bring them up to make sure they're double checked and can be ruled out as a possible cause to the problem. Once all user errors are ruled out, then it can be considered a programming fault. All technical support sides of companies go through the same type of measures. I hope you understand what I'm trying to say.

Do you have Follow AI Exactly Checked? And do you have any 3-4 plays anywhere in the AI?
 
geekor
offline
Link
 
Originally posted by cwisler
Mods have to ask that because it's general questioning before verifying something is a bug. Not everyone thinks about all the different situations, therefore they bring them up to make sure they're double checked and can be ruled out as a possible cause to the problem. Once all user errors are ruled out, then it can be considered a programming fault. All technical support sides of companies go through the same type of measures. I hope you understand what I'm trying to say.

Do you have Follow AI Exactly Checked? And do you have any 3-4 plays anywhere in the AI?


I understand, was hoping to skip the simlpe errors 9liek asking if the computer is plugged in).

Yes set to follow exactly, and I have a host of 3-4 plays as we only play 3-4 for 2 WR sets (we only even carry 2 NT's), and ahve 304 set as vase defense. Which is what I am thinking for some reason, even though the tester shows it should use my plays, that it reverted to the base defense, which is just F'd up
 
Captain Kaos
offline
Link
 
Originally posted by geekor
Originally posted by cwisler

Mods have to ask that because it's general questioning before verifying something is a bug. Not everyone thinks about all the different situations, therefore they bring them up to make sure they're double checked and can be ruled out as a possible cause to the problem. Once all user errors are ruled out, then it can be considered a programming fault. All technical support sides of companies go through the same type of measures. I hope you understand what I'm trying to say.

Do you have Follow AI Exactly Checked? And do you have any 3-4 plays anywhere in the AI?


I understand, was hoping to skip the simlpe errors 9liek asking if the computer is plugged in).

Yes set to follow exactly, and I have a host of 3-4 plays as we only play 3-4 for 2 WR sets (we only even carry 2 NT's), and ahve 304 set as vase defense. Which is what I am thinking for some reason, even though the tester shows it should use my plays, that it reverted to the base defense, which is just F'd up


Yeah, I'm curious as though maybe since something, not sure what, didn't match to follow the AI exactly -- it reverted back to basic settings. That'd be my guess. Either way, might be a good idea to do what Mat suggested -- I'm going to in a few minutes for the Wolfpack.

Just keep an eye on the wrong plays being called, see if it happens again. It may have just been one of those strange situations where it didn't calculate something right, therefore reverting to your basic settings.
 
geekor
offline
Link
 
Originally posted by cwisler
Yeah, I'm curious as though maybe since something, not sure what, didn't match to follow the AI exactly -- it reverted back to basic settings. That'd be my guess. Either way, might be a good idea to do what Mat suggested -- I'm going to in a few minutes for the Wolfpack.

Just keep an eye on the wrong plays being called, see if it happens again. It may have just been one of those strange situations where it didn't calculate something right, therefore reverting to your basic settings.


That's my point, it shouldn't do that... yes I'm here and I can make a catch as as Matt suggested. But how many others are out there without this knowledge who might lose games because it doesn't follow the AI exactly like it, and the test says, it should? That doesn't make it less of a bug, and quite frankly, cracks in the armor should be fixed BEFORE adding sparkles to it.
 
Captain Kaos
offline
Link
 
Unfortunately, I can't help you out much more. Most of my background knowledge is in C++, not Perl. I really wish I could, but without the proper background knowledge, I can't diagnose how and what calculations are made for each individual instance. I do know though, if games are run through multiple servers, some calculations may be skipped depending on how the code is written.

I'll see if I can get Crpyt, R8, or K in here to help you out a bit more. Once again, sorry.
 
tragula
title
offline
Link
 
Originally posted by geekor
That's my point, it shouldn't do that... yes I'm here and I can make a catch as as Matt suggested. But how many others are out there without this knowledge who might lose games because it doesn't follow the AI exactly like it, and the test says, it should? That doesn't make it less of a bug, and quite frankly, cracks in the armor should be fixed BEFORE adding sparkles to it.


Here is my OAI glitch
http://goallineblitz.com/game/replay.pl?game_id=735394&pbp_id=3279973

I think most OC/DC knows about a few per season.
 
tragula
title
offline
Link
 
 
r8
offline
Link
 
It really sucks, but this is something that we have never been able to track down. I have ran countless tests trying to force the AI into errors, but I've never been able to reproduce it consistently.

I've dug through people's AIs before and I've found instances where I absolutely beyond a shadow of a doubt thought that there was a real bug there, but I've never found a correlation between any of them, nor have I been able to reproduce any of those exact instances on the test server.

So, however much I hate to say it, Mat is right. The best option really is just to put a "catch-all" at the bottom and hope for the best.
 
tragula
title
offline
Link
 
Maybe sometime the QB/MLB misunderstand the call from the sideline and run the wrong play ?

One idea that will help is creating an automated script that will test each game to the AI and playbook (my guess that it will have to be on the server).
A huge database for the AI glitch may help to locate the source. Or at least it will help to gather reliable information how bad is this issue. If it is bad enough (I doubt that) it may be time for a real-time debugging with some tool that will mark the instance of the errors. If the problem is not too bad, then knowing how much errors is expected will help reducing flaming about this.
 
tomb4444
offline
Link
 
Originally posted by r8
It really sucks, but this is something that we have never been able to track down. I have ran countless tests trying to force the AI into errors, but I've never been able to reproduce it consistently.

I've dug through people's AIs before and I've found instances where I absolutely beyond a shadow of a doubt thought that there was a real bug there, but I've never found a correlation between any of them, nor have I been able to reproduce any of those exact instances on the test server.

So, however much I hate to say it, Mat is right. The best option really is just to put a "catch-all" at the bottom and hope for the best.


Have you ever tried to reproduce these various instances on the same server that they originally occurred on?
 
Page:
 


You are not logged in. Please log in if you want to post a reply.