Re: [Scidvspc-users] Crash when playing against UCI engine from setup position

[Ralph M. <ral...@ou...>]

Thanks Steve,

Adding the return to markExercise() at least prevents Scid vs. PC from crashing.

Cheers,
Ralph

-----Original Message-----
From: Steve A [mailto:ste...@gm...] 
Sent: 16 April 2015 12:58 PM
To: Ralph Möritz; scidvspc-users
Subject: Re: [Scidvspc-users] Crash when playing against UCI engine from setup position

Hmmm - Great stuff. There is one procedure i know crashes Scid, and this feature is using it.  It is engine::search. If anyone with engine programming experience would care to look at it ? Ralph's example seems a decent way to get it to dump core. Backtrace is below.

Until it is fixed, you can just add this return line to disable it in the uci game (tcl/tools/analysis.tcl) .....
proc markExercise { prevscore score } {
+ return
......


Cheers.


(gdb) bt
#0  0x0000000000481230 in Engine::Search(int, int, int, bool) ()
#1  0x0000000000481114 in Engine::Search(int, int, int, bool) ()
#2  0x0000000000480e62 in Engine::Search(int, int, int, bool) ()
#3  0x0000000000480e62 in Engine::Search(int, int, int, bool) ()
#4  0x0000000000480e62 in Engine::Search(int, int, int, bool) ()
#5  0x0000000000480e62 in Engine::Search(int, int, int, bool) ()
#6  0x0000000000480e62 in Engine::Search(int, int, int, bool) ()
#7  0x0000000000480e62 in Engine::Search(int, int, int, bool) ()
#8  0x000000000048190c in Engine::SearchRoot(int, int, int, MoveList*)
()
#9  0x0000000000481bef in Engine::Think(MoveList*) ()
#10 0x000000000041d1c3 in sc_pos_analyze(void*, Tcl_Interp*, int, char
const**) ()
#11 0x000000000041e9a9 in sc_pos(void*, Tcl_Interp*, int, char
const**) ()

On Thu, Apr 16, 2015 at 8:23 PM, Ralph Möritz <ral...@ou...> wrote:

> I’m using Scid vs. PC 4.14 x64 on Windows 8.1. I’m trying to practice 
> my endgame play by setting up a position, e.g. “8/8/8/6k1/1N6/8/8/K6B 
> w - - 0 1”, and playing against an engine. Unfortunately, after a few 
> moves by both sides, Scid vs. PC consistently crashes (the number of 
> moves before a crash seems to vary). I’ve only seen this happen when 
> playing against UCI engines, because Phalanx always resigns in lost endings J.