bug-gnubg
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Bug-gnubg] Pbs in current code (0.90) ?


From: Massimiliano Maini
Subject: Re: [Bug-gnubg] Pbs in current code (0.90) ?
Date: Wed, 9 Apr 2008 12:40:22 +0200


Michael Petch <address@hidden> wrote on 09/04/2008 12:11:50:


>> On 4/9/08 2:46 AM, "Massimiliano Maini" <address@hidden> wrote:
>>
>> Hi all,
>>
>> I've already reported an issue to the mailing list but nobody replied (maybe Christian said once that he
>> was unable to reproduice).
>>
>> I start a match (let's say 5pt) with tutor on (and analysis+evaluation at the same level, 2ply supremo).
>> At the 1st occasion, I double. Gnubg takes (obviously). Now I'm on roll and in the game record panel
>> I click back no my duoble decision to see the analysis: the analysis panle show something very strange.
>>
>> In some cases I get:
>>
>>         Cube analysis
>>         2-ply cubeless equity  +0.000 (Money:  -1.000)
>>          0.000 0.000 0.000 - 1.000 0.000 0.000
>>         Cubeful equities:
>>         1. No double            -6.456
>>         2. Double, pass         +1.000  ( +7.456)
>>         3. Double, take         -6.456  ( +0.000)
>>         Proper cube action: Optional double, take
>>
>> In some other cases I get some garbage in the L(g) percentage (something like 0.1234563257835838539829253).
>>
>> Notice that the hint works fine:
>>
>>         Cube analysis
>>         2-ply cubeless equity  -0.029 (Money:  -0.030)
>>          0.493 0.138 0.005 - 0.507 0.151 0.007
>>         Cubeful equities:
>>         1. No double            -0.039
>>         2. Double, pass         +1.000  ( +1.039)
>>         3. Double, take         -0.402  ( -0.363)
>>         Proper cube action: No double, take (25.9%)
>>
>> And if I clear this move analysis and re-analyse it, then it works fine.
>> Can anybody reproduce the problem ?
>>
>> Link to the exes you can use to test if you have the same problem (just put them in your install dir):
>>
>>        
http://www.gnubg.org/media/windows/gnubg-0.90-20080408-NOGLADE-exes-WMT.zip
>>
>>
>> Another pair of issues I've found (maybe it has been there forever):
>>
>> - start gnubg and load this position:
>>
>>                
http://www.gnubg.org/media/LGDD-pos.sgf
>>
>> - in order to roll, I have to use ctrl-R or the menu, if I just click in the middle area of the board
>> it says "You must roll the dice before moving pieces".
>> - I roll and complete the game (whichever outcome)
>> - if it goes on to the next game, just go back to the played game and run the analysis
>> - now something strange happens: the screen is no longer updated, major redraw issues minimizing
>> and restoring the window, trying to close gnubg it sometimes crashes badly.
>>
>> Again, can anybody reproduce ?
>>
>>
>> MaX.

>>
> I saw something similar to this after building a MAC OS/X release
> yesterday. The Luck factors has huge monster numbers in them after I
> had simply had the bot play itself an 11pt match on default settings
> then analyzed at Grandmaster/Supremo settings. I played another
> match afterwards and it was fine.
>
> I’m wondering if it might be related to some strange things I have
> noticed wiyh the latest .16 release for Windows (The multithreaded
> version published on the site).  I often will open a match, analyze
> it, save, open another math and then repeat. Often after analyzing a
> few matches the analysis will start showing weird results. In these
> cases I usually close down gnu, restart and do it over and its just
> fine. Sometimes if I run multiple analysis back to back (3 or 4
> manually) Gnu will crash, I then restart and all is fine. This
> morning I did a batch analysis (10 matches), and after the first 1
> or 2 matches it will crash. This wasn’t a problem with the stable
> release. Its almost as if state information between matches (or
> individual analysis isn’t cleared and it screws up a subsequent one)

Your problem seems more realted to a memory leak and/or some state not
being correctly cleared (as you suggested).

My tow ones are systematic and happens eraly on.
I tried with the single-thread exe and, when I go back on the missed
double decision the analysis panel shows some garbage (in thw W(g)
percentage). If I click on the copy button (text copy of the analysis
in the clipboard), it crashes. Single-thread exes are here:

        http://www.gnubg.org/media/windows/gnubg-0.90-20080408-NOGLADE-exes.zip

I also noticed something strange: when I experience redraw issues (board and menus
not refreshed, the cheque beside the player2 anme (in the lower part of the screen)
seems to appear at two or three slightly different locations (it moves horizontally
by a few pixels).

MaX.
reply via email to

[Prev in Thread] Current Thread [Next in Thread]