Two Plus Two Publishing LLC Two Plus Two Publishing LLC
 

Go Back   Two Plus Two Poker Forums > >

Free Software Discussion of Free / Freeware / Donationware / Open Source poker software and Free Graphics mods

Closed Thread
 
Thread Tools Display Modes
Old 07-09-2009, 10:14 AM   #851
jukofyork
Carpal \'Tunnel
 
jukofyork's Avatar
 
Join Date: Sep 2004
Location: Trollstopia, UK
Posts: 11,610
Re: FPHG available

Quote:
Originally Posted by jdp View Post
Is FPHG still being supported please? I'm getting a whole lot of "Invalid pot size.." errors at Party now.
It prolly related to the crash bug. It seems that having something reading the HH file while FPHG is writing it upsets Party on some PCs, so until I get it fixed it's not a good idea to use it with HEM or BetPot while you are playing if you are one of the affected people (there may be another bug related to writing corrupted HH file too, but it could also just be that it's finding party over-written ones in memory [the old FPHG used to find these too].

I can't say when I'll have time to work on it though or even if my idea will actually fix the problems (I have it about 50% written, but it still needs a couple of days to finish and test it), but I will try to add back the old method of memory scanning for people to fall back on if the crash bug isn't sorted out.

Juk
jukofyork is offline  
Old 07-28-2009, 02:41 AM   #852
epiLog
grinder
 
Join Date: Sep 2004
Posts: 491
Re: FPHG available

Just a bump to remind you of this. Would be awesome if you could fix this annoying crash bug
epiLog is offline  
Old 08-04-2009, 01:34 PM   #853
Pomtidom
grinder
 
Pomtidom's Avatar
 
Join Date: Sep 2006
Posts: 500
Re: FPHG available

Hey Juk,

I'm still having the (very annoying) issue I described here. And I've been thinking about a workaround that would help me out a lot. I came up with the following idea:

If I let FreePHG only write the observed handhistories, HEM will just pick up the handhistories of the Party client and I will not have this issue (OBSERVED_ONLY=1). I care much less about bugs in observed handhistories. The problem with this is that this setting applies to the .live files as well, so setting OBSERVED_ONLY to 1 won't allow me to use a BetPot script anymore.

Now what I would like to be able to do is let FreePHG output only observed handhistories but output .live handhistories of *all* tables. I did not see a way to create this scenario with the current options in freephg.ini. Is it possible for you to slightly tweak the FreePHG binary to exhibit this behavior? I'd be very grateful and am willing to compensate you for your time.

Let me know if anything isn't clear. Thanks!
Pomtidom is offline  
Old 08-04-2009, 03:18 PM   #854
jukofyork
Carpal \'Tunnel
 
jukofyork's Avatar
 
Join Date: Sep 2004
Location: Trollstopia, UK
Posts: 11,610
Re: FPHG available

Quote:
Originally Posted by epiLog View Post
Just a bump to remind you of this. Would be awesome if you could fix this annoying crash bug
Quote:
Originally Posted by Pomtidom View Post
Hey Juk,

I'm still having the (very annoying) issue I described here. And I've been thinking about a workaround that would help me out a lot. I came up with the following idea:

If I let FreePHG only write the observed handhistories, HEM will just pick up the handhistories of the Party client and I will not have this issue (OBSERVED_ONLY=1). I care much less about bugs in observed handhistories. The problem with this is that this setting applies to the .live files as well, so setting OBSERVED_ONLY to 1 won't allow me to use a BetPot script anymore.

Now what I would like to be able to do is let FreePHG output only observed handhistories but output .live handhistories of *all* tables. I did not see a way to create this scenario with the current options in freephg.ini. Is it possible for you to slightly tweak the FreePHG binary to exhibit this behavior? I'd be very grateful and am willing to compensate you for your time.

Let me know if anything isn't clear. Thanks!
I'll try and add something for this, but atm I'm just leaving the old code until I get time to finish the new code (which will be able to use both function hooking and memory scanning methods and also dump observed hands properly rather than waiting for and chopping off at the "Dealt to" line). I already wrote the IPC part, but have been busy with other stuff and havent had time to write the rest yet (no guarantee passing the work out will stop the crash bug, but at least there will be a fallback).

Juk
jukofyork is offline  
Old 08-04-2009, 05:07 PM   #855
jukofyork
Carpal \'Tunnel
 
jukofyork's Avatar
 
Join Date: Sep 2004
Location: Trollstopia, UK
Posts: 11,610
Re: FPHG available

I'll see if I can finish off the FPHG rewrite over the next 2-3 days. I was hoping to go away, but since the weather looks like it's gonna still be **** I'll have some free time on my hands. Will post again when I have more details.

Juk
jukofyork is offline  
Old 08-05-2009, 03:31 PM   #856
jukofyork
Carpal \'Tunnel
 
jukofyork's Avatar
 
Join Date: Sep 2004
Location: Trollstopia, UK
Posts: 11,610
Re: FPHG available

OK, I need some people (who suffer from the crash bug and/or get occasional corrupted hands) to test this: http://www.jukofyork.com/FreePHG_v3a.exe

NOTE: This is only a (very basic and cut-down) test version so don't bother downloading it if your current FPHG runs without any problems... I'll be releasing the proper FPHG V3 installation/EXE later on.

Just download the EXE and run it: It will create the data-mined ".hhf" hands in the "C:\FreePHG_HandHistories" folder and put the ".live" hands in the "C:\FreePHG_HandHistories\live" folder (no way to change this in this test version atm).

The code will be quite CPU intensive as it grabs the hands from memory like the old FPHG v0.0x versions did. This is what I intend to use as a fallback in case the hooking-related crash bug isn't solved (the hooking code will just use a different method of acquiring the data - all the same data-handling functions in this code will still be called so any crashes or corruption will clearly be the fault of this code and not the hooking part).

There will be a small delay between the actions of the players and the hand being written (as there will be with the new asynchronous hooking code) so make sure you don't act too quickly or BetPot might miss a player's action before you (there is a way to tune this so as to react quicker using more CPU time, but unless this becomes a problem for the testers I'm just going to leave it at 1000ms delay - post if this is a problem and/or it's using too much CPU to test properly and I'll tell you how to edit the command line arguments).

Here are the questions I need answering from those who test this:

A) Does this work with HEM and/or any other HUD that uses it?
B) Does this work with BetPot and if so is the 1000ms scanning delay OK?
C) Does this use too much CPU time and cause your client/system to chug? If so what CPU are you using?
D) Do any odd corrupted hands appear? These might be hands with random junk in them or hands where actions have been missed or delayed until future lines get written. If so, please post them.
E) Overall is this satisfactory to use if I cannot fix the hooking crash bug for the people who get it?


If people can try this and answer these questions then the next step is to try acquiring the data via hooking again, but instead of doing all the work inside of Party's memory (like I did before) pass out the data as soon as possible to then be passed through this same set of (crash and corruption tested) functions (running within FPHG.exe rather than Party itself). If Party still crashes then then there is nothing I can do about it, but hopefully if it does the fallback code will still be adequate for those who can't use the hooking.

I have the IPC code on my netbook and it's just a case of gluing it all together to create FPHG V3 after I get some feedback on this test version (a few hours work at most).

Juk

Last edited by jukofyork; 08-05-2009 at 03:37 PM.
jukofyork is offline  
Old 08-05-2009, 07:42 PM   #857
jukofyork
Carpal \'Tunnel
 
jukofyork's Avatar
 
Join Date: Sep 2004
Location: Trollstopia, UK
Posts: 11,610
Re: FPHG available

Nobody? It would be nice if somebody could test this if possible I'm not at all sure how well it'll work for slower/single-core CPUs even though it seems to run fine on mine (nobody has used FPHG 0.0x for about 3 years afaik and most just had 1 core back then...).

I'm now at the stage where I have Party passing out the HH data to the FPHG executable and all seems to be working fine (I've had 18 tables open and mining using this for the last hour without any problems - and it's even working synchronously atm).

I still need to add a buffer at the receiver end, but that's all that needs to be done. It turns out that if I run the "deal with hand data" loop as a separate thread (which just reads from the buffer) then I'll be able to put HH data into the buffer from either memory scanning or hooked IPC seamlessly and for people with multi-core CPUs it'll mean that memory scanning can be speeded up slightly (ie: one thread scanning for hands while another deals with and writes them out, etc).

Anyway, gonna call it a night now, but it should be finished by tomorrow afternoon/evening.

Juk
jukofyork is offline  
Old 08-06-2009, 04:04 AM   #858
Peter
adept
 
Join Date: Sep 2002
Posts: 889
Re: FPHG available

Juk, I'll give it a try today.
Peter is offline  
Old 08-06-2009, 04:21 AM   #859
gball
adept
 
gball's Avatar
 
Join Date: May 2006
Posts: 868
Re: FPHG available

I found a work-around for the freezing that works for me, when I notice party freezing I quickly kill hmhud.exe (the HEM hud process which seems to have gone into a busy loop), and then party starts responding again. Then I can start the hud again and it works ok until it happens again in 60 minutes or so. Could be a problem in the hud itself, but I'll still try the new version later today.
gball is offline  
Old 08-06-2009, 06:04 AM   #860
jukofyork
Carpal \'Tunnel
 
jukofyork's Avatar
 
Join Date: Sep 2004
Location: Trollstopia, UK
Posts: 11,610
Re: FPHG available

Quote:
Originally Posted by gball View Post
I found a work-around for the freezing that works for me, when I notice party freezing I quickly kill hmhud.exe (the HEM hud process which seems to have gone into a busy loop), and then party starts responding again. Then I can start the hud again and it works ok until it happens again in 60 minutes or so. Could be a problem in the hud itself, but I'll still try the new version later today.
This sounds like HEM is somehow locking the file which will have upset Party quite badly if it did - Party has some kind of "freeze sensing" which restarts itself.

Basically the whole idea of the new code is to get all of the file writing stuff and all of the unnecessarily complex code out of Party's memory space so that HEM/BetPot file locking and/or a bug can't upset Party anymore.

I'll have the first version of FPHG v3 out in a few hours and hopefully I can fix any minor issues with the new code brought up here by the end of the weekend (I think I'm gonna be going away at the beginning of next week).

Juk
jukofyork is offline  
Old 08-06-2009, 12:17 PM   #861
jukofyork
Carpal \'Tunnel
 
jukofyork's Avatar
 
Join Date: Sep 2004
Location: Trollstopia, UK
Posts: 11,610
Post FPHG 3.00 BETA available

OK, here is the first BETA test version of FPHG v3.00: http://www.jukofyork.com/FreePHG_Install_BETA.exe

Just download and install as you did with FPHG v2.0x - The format of the 'FreePHG.ini' file is exactly the same as before, but there are a few small changes/improvements to the user functionality (see below).

I haven't added the memory scanning fallback code yet as I've not had any feedback on whether it is satisfactory or not yet and also if the changes in this version fix the crash bug there will be no need for it (ie: Hooking >>> Memory Scanning). If it does turn out that it's needed then it's trivial to add to the new code (will need a couple more options adding to the INI to tune it though).

SYSTEM FUNCTIONALITY CHANGES FROM FreePHG V2.0x
  • The code injected into Party's memory space now does the absolute bare minimum processing and then sends the raw data (via 'WM_COPYDATA' IPC code) to the FPHG process for processing.
  • No files are opened or written to from within Party's memory space now.
  • The hand processing within then FPHG process runs in a separate thread so can take advantage of machines with multiple cores (hardly makes any difference to the hook-based code, but will make a big difference for the memory scanning version if needed).

USER FUNCTIONALITY CHANGES FROM FreePHG V2.0x
  • The new default location for storing hands is now just below the folder where FPHG is installed to (ie: ".\HandHistories"). It will still read the old v2.0x FreePHG.ini file and store to "C:\FreePHG_HandHistories", etc if it finds it.
  • If the OBSERVED_ONLY flag is set then FPHG should no longer write out erroneous "hand headers" (ie: text up to the "** Dealing down cards **" line) for non-observed hands. It now waits to see what gets written after the "** Dealing down cards **" line and only starts writting the hand out if it's not the "Dealt to" line.
  • If the OUTPUT_LIVE flag is not set then FPHG will no longer bother to create the "live" subfolder and will also only write out fully complete hands (ie: it won't write them out in real-time after every action, etc). This saves thrashing the hard disk when we don't care about the hands being written in realtime. This should also fix the problem of trackers reading partial hands (there may still be a tiny problem with side-pots getting written as an extra bit "tacked on" - I'll try and fix this in a later version).
  • If the OUTPUT_LIVE flag is set the live subfolder will be created and removed on the fly as FPHG starts up and shuts down. This will mean that the folder won't be visible to "browse to" in HEM until after you start FPHG with the flag set.

Please report back how you get on with this. Does it still cause Party to crash? Is the OBSERVED_ONLY mode working better? Is it working as expected with the OUTPUT_LIVE flag on and off (see above). Is there anything else I've messed up (not playing poker atm so only tested briefly on play money tables)?

Juk

Last edited by jukofyork; 08-06-2009 at 03:01 PM. Reason: Clarified new "OBSERVED_ONLY" flag functionality and bolded other important changes in user functionality.
jukofyork is offline  
Old 08-06-2009, 03:10 PM   #862
gball
adept
 
gball's Avatar
 
Join Date: May 2006
Posts: 868
Re: FPHG 3.00 BETA available

Quote:
Originally Posted by jukofyork View Post
SYSTEM FUNCTIONALITY CHANGES FROM FreePHG V2.0x[LIST][*] The code injected into Party's memory space now does the absolute bare minimum processing and then sends the raw data (via 'WM_COPYDATA' IPC code) to the FPHG process for processing.
Haven't really tested it yet, just got it installed, but now as you are already working on the program, could you make it possible to define additional process(es) in the .ini file where the data would also be sent? This would make a bit more "cleaner" design possible for any 3rd party bet pot tool, because they wouldn't have to poll and scan the fphg output files. What do you think?
gball is offline  
Old 08-06-2009, 03:21 PM   #863
jay-pee
stranger
 
Join Date: Dec 2007
Posts: 1
Re: FPHG available

I'm still picking up some anomalies in the handhistories, some examples:
Code:
Game #8308415312 starts.

#Game No : 8308415312 
***** Hand History for Game 8308415312 *****
$2 USD NL Texas Hold'em - Thursday, August 06, 15:04:57 EDT 2009
Table Table  169497 (Real Money)
Seat 3 is the button
Total number of players : 9 
Seat 9: FerrumVS ( $1.54 USD )
Seat 4: Liliah ( $0.40 USD )
Seat 8: Lonlilokli1 ( $4.16 USD )
Seat 7: SlowBlues ( $1.88 USD )
Seat 6: The_Hague ( $1.74 USD )
Seat 1: boroka333 ( $2.45 USD )
Seat 2: jasiu82 ( $0.37 USD )
Seat 5: noddy101 ( $0.98 USD )
Seat 3: z22036th ( $2.19 USD )
noddy101 posts small blind [$0.01 USD].
The_Hague posts big blind [$0.02 USD].
** Dealing down cards **
SlowBlues folds
Lonlilokli1 folds
FerrumVS folds
jasiu82 folds
z22036th folds
noddy101 calls [$0.01 USD]
The_Hague checks
** Dealing Flop ** [ 6h, 4d, 6d ]
noddy101 checks
The_Hague bets [$0.08 USD]
noddy101 folds
cards.
ue does not show cards.
The_Hague wins $0.12 USD

Game #8308417917 starts.

#Game No : 8308417917 
***** Hand History for Game 8308417917 *****
$2 USD NL Texas Hold'em - Thursday, August 06, 15:05:40 EDT 2009
Table Table  169497 (Real Money)
Seat 5 is the button
Total number of players : 9 
Seat 9: FerrumVS ( $1.54 USD )
Seat 4: Liliah ( $0.40 USD )
Seat 8: Lonlilokli1 ( $4.16 USD )
Seat 7: SlowBlues ( $1.88 USD )
Seat 6: The_Hague ( $1.76 USD )
Seat 1: boroka333 ( $2.45 USD )
Seat 2: jasiu82 ( $0.37 USD )
Seat 5: noddy101 ( $0.96 USD )
Seat 3: z22036th ( $2.19 USD )
The_Hague posts small blind [$0.01 USD].
SlowBlues posts big blind [$0.02 USD].
** Dealing down cards **
Lonlilokli1 raises [$0.08 USD]
FerrumVS folds
jasiu82 folds
z22036th folds
noddy101 folds
The_Hague calls [$0.07 USD]
SlowBlues folds
** Dealing Flop ** [ 2c, 2d, 7d ]
The_Hague bets [$0.10 USD]
Lonlilokli1 calls [$0.10 USD]
** Dealing Turn ** [ Ac ]
The_Hague bets [$0.40 USD]
Lonlilokli1 calls [$0.40 USD]
** Dealing River ** [ 6h ]
The_Hague is all-In  [$1.18 USD]
d in time.(disconnected)
pond in time.(disconnected)
Lonlilokli1 folds

he_Hague does not show cards.
The_Hague wins $2.31 USD

Game #8308382832 starts.

#Game No : 8308382832 
***** Hand History for Game 8308382832 *****
$2 USD NL Texas Hold'em - Thursday, August 06, 14:54:37 EDT 2009
Table Table  169623 (Real Money)
Seat 8 is the button
Total number of players : 8 
Seat 7: Orbas123 ( $1.96 USD )
Seat 2: PharoahG ( $1.36 USD )
Seat 9: Zembonator ( $2.75 USD )
Seat 4: joaoZa ( $2.67 USD )
Seat 1: kabil888 ( $0.83 USD )
Seat 3: napaalm ( $2 USD )
Seat 8: nuuunja ( $0.46 USD )
Seat 6: twaxxo ( $2 USD )
Zembonator posts small blind [$0.01 USD].
osts big blind [$0.02 USD].
.02 USD].
** Dealing down cards **
PharoahG folds
napaalm folds
twaxxo calls [$0.02 USD]
Orbas123 folds
nuuunja folds
Zembonator calls [$0.01 USD]
kalambaha has joined the table.
kabil888 checks
** Dealing Flop ** [ 5s, Qd, Ad ]
Zembonator checks
kabil888 bets [$0.02 USD]
twaxxo calls [$0.02 USD]
Zembonator calls [$0.02 USD]
** Dealing Turn ** [ 9c ]
Zembonator checks
kabil888 bets [$0.02 USD]
twaxxo raises [$0.13 USD]
Zembonator calls [$0.13 USD]
kabil888 calls [$0.11 USD]
** Dealing River ** [ Ah ]
Zembonator checks
kabil888 checks
twaxxo bets [$0.36 USD]
Zembonator calls [$0.36 USD]
kabil888 calls [$0.36 USD]
twaxxo shows [ 9h, 9d ]a full house, Nines full of Aces.
Zembonator doesn't show [ 4c, 5h ]two pairs, Aces and Fives.
kabil888 doesn't show [ 7d, Qh ]two pairs, Aces and Queens.
twaxxo wins $1.52 USD from the main pot with a full house, Nines full of Aces.
jay-pee is offline  
Old 08-06-2009, 03:41 PM   #864
jukofyork
Carpal \'Tunnel
 
jukofyork's Avatar
 
Join Date: Sep 2004
Location: Trollstopia, UK
Posts: 11,610
Re: FPHG 3.00 BETA available

Quote:
Originally Posted by gball View Post
Haven't really tested it yet, just got it installed, but now as you are already working on the program, could you make it possible to define additional process(es) in the .ini file where the data would also be sent? This would make a bit more "cleaner" design possible for any 3rd party bet pot tool, because they wouldn't have to poll and scan the fphg output files. What do you think?
Yeah this would be possible and I just checked the AHK forums and there is a code snippet that shows how to recieve the data here.

The big problem with this idea is the fact that the HH data being sent hasn't been processed (as it's the FPHG process that does all the processing) and you would end up having to do all of the same book-keeping as FPHG does.

There is no reason why the FPHG process couldn't transmit the processed hand history data to other processes using the same method, but I'm not 100% sure it wouldn't actually end up harder than just reading the file when a decision is required (do any applications actually poll the FPHG files?).

Juk
jukofyork is offline  
Old 08-06-2009, 03:47 PM   #865
jukofyork
Carpal \'Tunnel
 
jukofyork's Avatar
 
Join Date: Sep 2004
Location: Trollstopia, UK
Posts: 11,610
Re: FPHG available

Quote:
Originally Posted by jay-pee View Post
I'm still picking up some anomalies in the handhistories, some examples:
Code:
Game #8308415312 starts.

#Game No : 8308415312 
***** Hand History for Game 8308415312 *****
$2 USD NL Texas Hold'em - Thursday, August 06, 15:04:57 EDT 2009
Table Table  169497 (Real Money)
Seat 3 is the button
Total number of players : 9 
Seat 9: FerrumVS ( $1.54 USD )
Seat 4: Liliah ( $0.40 USD )
Seat 8: Lonlilokli1 ( $4.16 USD )
Seat 7: SlowBlues ( $1.88 USD )
Seat 6: The_Hague ( $1.74 USD )
Seat 1: boroka333 ( $2.45 USD )
Seat 2: jasiu82 ( $0.37 USD )
Seat 5: noddy101 ( $0.98 USD )
Seat 3: z22036th ( $2.19 USD )
noddy101 posts small blind [$0.01 USD].
The_Hague posts big blind [$0.02 USD].
** Dealing down cards **
SlowBlues folds
Lonlilokli1 folds
FerrumVS folds
jasiu82 folds
z22036th folds
noddy101 calls [$0.01 USD]
The_Hague checks
** Dealing Flop ** [ 6h, 4d, 6d ]
noddy101 checks
The_Hague bets [$0.08 USD]
noddy101 folds
cards.
ue does not show cards.
The_Hague wins $0.12 USD

Game #8308417917 starts.

#Game No : 8308417917 
***** Hand History for Game 8308417917 *****
$2 USD NL Texas Hold'em - Thursday, August 06, 15:05:40 EDT 2009
Table Table  169497 (Real Money)
Seat 5 is the button
Total number of players : 9 
Seat 9: FerrumVS ( $1.54 USD )
Seat 4: Liliah ( $0.40 USD )
Seat 8: Lonlilokli1 ( $4.16 USD )
Seat 7: SlowBlues ( $1.88 USD )
Seat 6: The_Hague ( $1.76 USD )
Seat 1: boroka333 ( $2.45 USD )
Seat 2: jasiu82 ( $0.37 USD )
Seat 5: noddy101 ( $0.96 USD )
Seat 3: z22036th ( $2.19 USD )
The_Hague posts small blind [$0.01 USD].
SlowBlues posts big blind [$0.02 USD].
** Dealing down cards **
Lonlilokli1 raises [$0.08 USD]
FerrumVS folds
jasiu82 folds
z22036th folds
noddy101 folds
The_Hague calls [$0.07 USD]
SlowBlues folds
** Dealing Flop ** [ 2c, 2d, 7d ]
The_Hague bets [$0.10 USD]
Lonlilokli1 calls [$0.10 USD]
** Dealing Turn ** [ Ac ]
The_Hague bets [$0.40 USD]
Lonlilokli1 calls [$0.40 USD]
** Dealing River ** [ 6h ]
The_Hague is all-In  [$1.18 USD]
d in time.(disconnected)
pond in time.(disconnected)
Lonlilokli1 folds

he_Hague does not show cards.
The_Hague wins $2.31 USD

Game #8308382832 starts.

#Game No : 8308382832 
***** Hand History for Game 8308382832 *****
$2 USD NL Texas Hold'em - Thursday, August 06, 14:54:37 EDT 2009
Table Table  169623 (Real Money)
Seat 8 is the button
Total number of players : 8 
Seat 7: Orbas123 ( $1.96 USD )
Seat 2: PharoahG ( $1.36 USD )
Seat 9: Zembonator ( $2.75 USD )
Seat 4: joaoZa ( $2.67 USD )
Seat 1: kabil888 ( $0.83 USD )
Seat 3: napaalm ( $2 USD )
Seat 8: nuuunja ( $0.46 USD )
Seat 6: twaxxo ( $2 USD )
Zembonator posts small blind [$0.01 USD].
osts big blind [$0.02 USD].
.02 USD].
** Dealing down cards **
PharoahG folds
napaalm folds
twaxxo calls [$0.02 USD]
Orbas123 folds
nuuunja folds
Zembonator calls [$0.01 USD]
kalambaha has joined the table.
kabil888 checks
** Dealing Flop ** [ 5s, Qd, Ad ]
Zembonator checks
kabil888 bets [$0.02 USD]
twaxxo calls [$0.02 USD]
Zembonator calls [$0.02 USD]
** Dealing Turn ** [ 9c ]
Zembonator checks
kabil888 bets [$0.02 USD]
twaxxo raises [$0.13 USD]
Zembonator calls [$0.13 USD]
kabil888 calls [$0.11 USD]
** Dealing River ** [ Ah ]
Zembonator checks
kabil888 checks
twaxxo bets [$0.36 USD]
Zembonator calls [$0.36 USD]
kabil888 calls [$0.36 USD]
twaxxo shows [ 9h, 9d ]a full house, Nines full of Aces.
Zembonator doesn't show [ 4c, 5h ]two pairs, Aces and Fives.
kabil888 doesn't show [ 7d, Qh ]two pairs, Aces and Queens.
twaxxo wins $1.52 USD from the main pot with a full house, Nines full of Aces.
Yeah, I think whatever I do there is always going to be a few oddball hand histories like this (the memory scanning code is most prone to this). I've just tried to add something to v3.01 to try to reduce it a bit, but I don't think it'll ever work 100% (luckily they tend to occur at the end hands most often so are less likly to screw up BetPot and HEM).

The problem is that you are fishing hands out of memory where new hands write over the top of old ones and bits of hands get copied about and they can end up "looking like" valid hand histories. The only surefire solution to this would be to parse every hand history you find and strictly reject any that don't pass the parser, but then you would get the opposite problem where any tiny change in the hand history format or addition of a new game would cause hands to get rejected (not to mention the added work of maintaining it all).

Juk
jukofyork is offline  
Old 08-06-2009, 04:25 PM   #866
Peter
adept
 
Join Date: Sep 2002
Posts: 889
Re: FPHG available

Juk, do you still need the FreePHG_v3a to be tested? I didn't get to play today, but can try on sunday.
Peter is offline  
Old 08-06-2009, 04:32 PM   #867
jukofyork
Carpal \'Tunnel
 
jukofyork's Avatar
 
Join Date: Sep 2004
Location: Trollstopia, UK
Posts: 11,610
Re: FPHG available

Quote:
Originally Posted by Peter View Post
Juk, do you still need the FreePHG_v3a to be tested? I didn't get to play today, but can try on sunday.
Only if the new FPHG v3.00 code doesn't stop the crash-bug (and if it doesn't then I'll add the code to the real FPHG v3.0x code which will be easier to use than the FreePHG_v3a test exe). The main priority atm is to see if the hooking-related crash-bug is fixed.

Juk
jukofyork is offline  
Old 08-06-2009, 06:41 PM   #868
gball
adept
 
gball's Avatar
 
Join Date: May 2006
Posts: 868
Re: FPHG available

New version didn't help the hem hud freezing issue.
gball is offline  
Old 08-06-2009, 07:09 PM   #869
jukofyork
Carpal \'Tunnel
 
jukofyork's Avatar
 
Join Date: Sep 2004
Location: Trollstopia, UK
Posts: 11,610
Re: FPHG available

Quote:
Originally Posted by gball View Post
New version didn't help the hem hud freezing issue.
It's just struck me that the HEM HUD causing the Party client to freeze up probably isn't anything to do with FPHG. If the HEM HUD uses a trick like using SetParent() to attach it's HUD window(s) to the tables then when the HEM HUD process hangs it will also cause the Party process to freeze as all of the message queues will stop working.

I've actually seen this happen myself when I tested different methods of attaching HUD windows to the tables a few years ago and the fact you say that killing the HEM HUD process quickly stops the Party client from killing itself reminds me of this too (ie: you used to get about 4-5 seconds before the Party client would think it's frozen and try to restart itself).

It may still be that something is happening between FPHG's output files and HEM (ie: some "war" over locked files or something), but it's not really the fault of FPHG if the crash occurs inside of HEM and is caused by HEM reading the FPHG files. If this is the case them it might be a good idea for HEM to first make a temporary copy (using OS calls like C's system("copy xxx yyy") command) of a file it wants to read.

There is a different "crash bug" that is different to this HEM HUD one. The people who got this were the same people got strange graphical and UI "glitches" in the client after a certain update in Jan/Fen this year, but then the same problem seemed to turn into the "crash bug" later on in the year.

Peter who posted earlier reported that even FPHG on it's own could cause it and I'm pretty sure others who only use FPHG with BetPot have also had this.

Juk
jukofyork is offline  
Old 08-06-2009, 07:31 PM   #870
jukofyork
Carpal \'Tunnel
 
jukofyork's Avatar
 
Join Date: Sep 2004
Location: Trollstopia, UK
Posts: 11,610
Re: FPHG available

I just had a quick hunt on the HEM forums and found a post in this thread about the HEM hanging issue:

Quote:
There logs I got from Bezille seemed to indicate his problem was confined to live tracking. These problem happen because a recent Party update causes FPHG to output mangled lines to the live folder. Going back to 1.7 wouldnt fix this.
So it looks like it's corrupted output from FPHG that is causing HEM HUD to lock up and that in turn is causing the Party client to freeze and then try to kill/restart itself (ie: in the way I mentioned above).

There is very little I can do to help fix this as the corrupted output is just a side effect of fishing hands out of memory (it must just be that a recent Party update caused HH data to corrupt in a way that really upsets HEM HUD when it reads it).

Another option is to fish about more inside of Party and see if I can find a better function to hook which is only called once per piece of hand data (the current function copies the same bits about quite a bit and sometimes even copies bits of random memory about that happen to have hands in).

I'm still interested in hearing from people who get the crash bug that is unrelated to the HEM HUD (ie: those who don't use HEM HUD or crashes that aren't preceded by HEM HUD locking up and freezing the Party client).

Juk
jukofyork is offline  
Old 08-07-2009, 07:28 AM   #871
jukofyork
Carpal \'Tunnel
 
jukofyork's Avatar
 
Join Date: Sep 2004
Location: Trollstopia, UK
Posts: 11,610
Post FPHG 3.02 BETA available

New BETA test version of FPHG v3.02: http://www.jukofyork.com/FreePHG_Install_BETA.exe

The changes in this version concentrate on trying to avoid writing corrupted hand data as much as possible. I think that at least some of the corrupted hands were to do with the "this player will use his time bank" and "this player has been disconnected" type message, but I'm not 100% sure sure.

Be on the lookout for (and post here) hands that still have corrupted data in them. If possible, please post the "real" hands history generated by Party to compare with (ie: for non-observed hands) as this helps me see any text that is getting delayed (ie: delayed output gives other stuff chance to write junk over the hand).

Also since I've made some of the matching even more strict there is a possibility that some valid hands will get rejected - be on the look out for any hands which just get totally ignored by FPHG (again try to post the "real" hands history when possible).

I still haven't heard from anybody who used to get the crash-bug whether the new method in V3.0x has actually fixed it or not so still interested in hearing about that (note though that the HEM HUD hanging thing is not the same as the crash-bug - see the above 2-3 posts).

NOTE: General info about the changes made in FPHG V3 can be found in this post.

Juk
jukofyork is offline  
Old 08-08-2009, 09:24 AM   #872
jukofyork
Carpal \'Tunnel
 
jukofyork's Avatar
 
Join Date: Sep 2004
Location: Trollstopia, UK
Posts: 11,610
Post FPHG 3.03 BETA available

New BETA test version of FPHG v3.03: http://www.jukofyork.com/FreePHG_Install_BETA.exe

The main change in this version is the addition of a "fallback" memory scanning mode for those who still suffer from Party client crashes:

"Passive Mode" (default)
In this mode the code works just like it always did (ie: it will grab the hands from Party using an injected DLL). Unless you suffer from Party client crashes and/or Party client "glitches" then you should stick to using passive mode as it uses much less CPU time and writes the hands out almost instantaneously.

"Active Mode"
In this mode the code will actively scan Party's memory for hands (ie: no DLL is injected so there is no possibility of crashing the Party client). To enable active mode then you need to open up "FreePHG.ini" and set the USE_ACTIVE_MODE flag to 1. The default scanning delay and thread priority should be fine for most users, but if you wish to only use FPHG for background data-mining then setting the ACTIVE_DELAY value to 3000 or 5000 would mean it uses a lot less CPU time. On the other hand if you have a fast CPU and/or a multi-core CPU and use FPHG for BetPot (or anything else that needs the output updated very quickly after each player's action) then you might want to try setting the ACTIVE_DELAY value to 300 or 500.

NOTE: Custom settings in the "FreePHG.ini" will need to be re-set after installing this version as your old "FreePHG.ini" will be overwritten by the installer.

NOTE: General info about the changes made in FPHG V3 can be found in this post.

Juk

PS: I will update the main FPHG page and change the installer on that page as soon as I get some feedback on v3.03 (I'm not planning on any more major changes to FPHG now and just need to know if all the changes I have made are working OK and also if the default settings for the new "Active Mode" are OK for most people).

Last edited by jukofyork; 08-08-2009 at 09:35 AM.
jukofyork is offline  
Old 08-08-2009, 10:15 AM   #873
jukofyork
Carpal \'Tunnel
 
jukofyork's Avatar
 
Join Date: Sep 2004
Location: Trollstopia, UK
Posts: 11,610
Re: FPHG 3.03 BETA available

Quote:
Originally Posted by jukofyork View Post
New BETA test version of FPHG v3.03: http://www.jukofyork.com/FreePHG_Install_BETA.exe
If you just downloaded this in the last hour and it wasn't saving any hands, then you need to re-download it (it had v3.02's DLL in the installer by accident - fixed now).

Juk
jukofyork is offline  
Old 08-09-2009, 05:35 AM   #874
Peter
adept
 
Join Date: Sep 2002
Posts: 889
Re: FPHG available

Played for awhile with the FreePHG_v3a version. Has the same problem and solution as has been mentioned before. The first two times it froze I killed HM Hud immediately and it was solved. The third time I waited for about half a minute, nothing happened and then when I killed HM Hud everything was fine again, except I had missing buttons on one table.
Playing without HM Hud seemed to give no problems and it also seemed to react better. With the Hud it was kind of slow with pot sized bets, UTG it even often just made it 0.00 and if I then waited and tried again it still didn't work. Without HM Hud I didn't seem to have this problem.

Peter
Peter is offline  
Old 08-09-2009, 06:29 AM   #875
jukofyork
Carpal \'Tunnel
 
jukofyork's Avatar
 
Join Date: Sep 2004
Location: Trollstopia, UK
Posts: 11,610
Re: FPHG available

Quote:
Originally Posted by Peter View Post
Played for awhile with the FreePHG_v3a version. Has the same problem and solution as has been mentioned before. The first two times it froze I killed HM Hud immediately and it was solved. The third time I waited for about half a minute, nothing happened and then when I killed HM Hud everything was fine again, except I had missing buttons on one table.
Playing without HM Hud seemed to give no problems and it also seemed to react better. With the Hud it was kind of slow with pot sized bets, UTG it even often just made it 0.00 and if I then waited and tried again it still didn't work. Without HM Hud I didn't seem to have this problem.
Hi Peter,

Can you try FPHG v3.03 (see above) to see if that works any better? It does more strict sanity checks than earlier versions so should be less likely to pass through corrupted data which HEM HUD doesn't like.

If possible try the first session with USE_ACTIVE_MODE left at 0 as this will work much better overall (but only if the crash bug has been fixed).

If this still causes the crash bug (ie: the one unrelated to HEM HUD) on your PC then you'll have to set USE_ACTIVE_MODE to 1 which will mean it scans memory like FreePHG_v3a did (except it has more strict sanity checking).

If this proves to be unresponsive you can then try to tune the ACTIVE_DELAY value down to say 500 and also try setting the ACTIVE_PRIORITY value to 0 or 1.

Juk
jukofyork is offline  

Closed Thread
      

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off


Forum Jump


All times are GMT -4. The time now is 04:45 AM.


Powered by vBulletin®
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.
Copyright © 2008-2020, Two Plus Two Interactive