Author: Jens Kahlenberg
Date: 10:14:47 06/29/03
Go up one level in this thread
On June 29, 2003 at 12:45:50, Dieter Buerssner wrote: >You could just print all global values to a log file at interesting times (for >example before/after each new/setboard). Then use a tool like diff. reproducable logs and diff ... that's everything one needs for regression testing: an efficient way to detect unwanted side effects if you've just made some changes in some desert area of the source (not affecting the log) thanx again for the link, jaxx > >Don't forget to clear history/killer tables. You might also want to print things >like piece_lists. > >Make sure, that this really is a bug :-) I found a similar "no bug" one time. >Actually Yace should always be able to reproduce exact nodes. But the following >happened. One run gave more output than the other (I did not care first). >Displaying moves can reorder the piece_list (for SAN output, I need to make >moves on the board, and undo them later. This can change the exact layout of the >piece lists. And when everything is the same, moveordering will use earlier >generated moves earlier than later generated ones. And moves are generated in >the order of the pieces like they appear in the piece lists). > >Now, it happened, that in one run I set the noise level different, than in the >other run. It would be more subtle, when the "noise level" would actually be a >time - say display PVs, when they are for searches of more than one second. > >Regards, >Dieter
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.