| commit | d4dcf61c68a94ebc18d8d2fc771c27751cd085d5 | |
| author | H.G.Muller <hgm@hgm-xboard.(none)> | |
| Fri, 21 Dec 2018 13:40:26 +0000 (21 14:40 +0100) | ||
| committer | H.G.Muller <hgm@hgm-xboard.(none)> | |
| Fri, 21 Dec 2018 15:21:37 +0000 (21 16:21 +0100) | ||
| tree | 99a5f9e50387de3774f556f44133226d4849adfc | treesnapshot (tar.gz zip) |
| parent | 1790aaa2dbb4c327b5e867e3e5c58547f6fd2ec3 | commitdiff |
Toggle stm in engine thread
Changing the side to move out of sync (i.e. in the GUI thread) sometimes
led to starting of a spurious search, when a 'force' command preceding
the moves was still queued. We now have the dummy command that was there
to release the engine thread from waiting for the queue also toggle stm.
This required a slightly different logic for handling ponder hits in
the GUI thread. These can still safely toggle stm there, as the engine
won't stop pondering before it gets the ponder hit.
To make sure incoming empty lines will not wreck this scheme, these
are now ignored.
Changing the side to move out of sync (i.e. in the GUI thread) sometimes
led to starting of a spurious search, when a 'force' command preceding
the moves was still queued. We now have the dummy command that was there
to release the engine thread from waiting for the queue also toggle stm.
This required a slightly different logic for handling ponder hits in
the GUI thread. These can still safely toggle stm there, as the engine
won't stop pondering before it gets the ponder hit.
To make sure incoming empty lines will not wreck this scheme, these
are now ignored.
| UCI2WB.c | diffblobblamehistory |