Scid 3.6.13 released

[pgeorges]

I uploaded a Scid 3.6.14rc1 version that better handles engines.
Just tell me if you still encounter some problems.

[rkp]

I tried multi pv. If 2 or greater, I get analysis. If 1, Rybka doesn't work. This was not the case with 3.6.11 or lower.

[rkp]

I've tried Scid 3.6.14rc2 and Rybka is working again. Thank you!

[pgeorges]

Did you try the latest RC of Scid 3.6.14 ?

[royb]

I can get Rybka-2.2 to work with this version. However the configure UCI engine dialog allows you to specify the hash table size. No matter what I enter in there (the default is 32 MB and I specify 256 MB. The result is that this setting is ignored.

This is not a Rybka-specific issue. Using Fruit-2.1 (free version) I get the same behavior.

I'm verifying the memory usage by issuing 'free -tm' and observing the memory consumed. 'top' also shows a trivial amount of memory used by the enigne when running in analysis mode.

Am I missing something about configuring UCI engines under SCID or is this a bug?

Thanks for your improvements! SCID is easily the best chess database for Linux now!

Roy

[pgeorges]

I uploaded a new rc3 of Scid 3.6.14 that seems to correct this bug : some UCI engines don't obey orders given consecutively : so I wrapped every option with "isready" and "readyok".
I tested it with Toga and Arasan under Linux and Rybka under Windows (Rybka memory comsumption under wine is unreliable).

[royb]

That nailed it!

Thanks for the quick fix to this!

Roy

[royb]

I think I've found another bug in 3.6.14-rc3 (could well be it has existed for several versions).

Start an engine in analysis mode. The output of the engine scrolls by as the depth of search increases. Now, drag the elevator button on that analysis window to scroll back to see ply 10, 9, 8, etc. You cannot see the output. It shows the same (current) ply output over and over.

The same is true if you expand the window downward to reveal previous ply's PVs -- they all show as the same (current) PV.

[pgeorges]

I think I've found another bug in 3.6.14-rc3 (could well be it has existed for several versions).

Start an engine in analysis mode. The output of the engine scrolls by as the depth of search increases. Now, drag the elevator button on that analysis window to scroll back to see ply 10, 9, 8, etc. You cannot see the output. It shows the same (current) ply output over and over.

This is not a bug : it is a feature
When you scroll back, you will see previous lines but as soon as the engine sends new output you will be automatically sent back to the bottom of the lines (the scrollbar goes to the end).

The same is true if you expand the window downward to reveal previous ply's PVs -- they all show as the same (current) PV.

I can't reproduce this : can you precise what engine you use (xboard, UCI) and the number of Lines (multi-pv) ?