Seeing that the Elo graph in tournament tab has labels showing the value of each point when you hover over it, I wonder if it can also be done for the graphs in the game tab. It would be very useful I believe.
Also, I don't know if it's just me, but I'm used to being able to click on a point in the eval graph to jump to that move on the board. Is that something you'd consider adding?
Max wrote: ↑Thu Sep 10, 2026 10:22 am
On Linux, a clipboard processing issue occurs when copying text in another application to paste it into Myracle.
Example: Open the info window of an installed engine in Tools/Settings. Then copy (Ctrl-C) text in another app such as Firefox and paste it (Ctrl-V) into Myracle. Most of the time nothing happens, but sometimes an older, previously copied text is pasted from the clipboard.
I've uploaded new linux zip files, could you try it please? still weird in WSL but seems to work in Ubuntu in my virtual Linux machine.
looks like some apps use legaci iso-latin-1 encoding and X11 fails to convert to utf8 internally so I'm trying to handle both now
build 122 is up:
- limit swiss rounds to (P+1)/2, otherwise sometimes last rounds would fail to pair
- fix cutechess engine import (split command part into executable + arguments)
- add simple per-engine tournament stats (stats button in tournament tab, to the left of graph button, toggles crosstable/stats view)
- mouse hover now also works for standard graphs, clicking the graph navigates to linked move
build 123 is up:
- critical fix: SAN parser accepts en-passant captures out of check; something I broke in build 121
(thanks to Fonzy for reporting this)
I urge anyone using builds 121 or 122 to update to 123 as soon as possible, my apologies for this
- cosmetic: move time in tournament stats has more resoltion instead of being quantized to seconds
I'll try to describe what was impacted and what was not and why:
I made some perf optimization of the SAN parser (in build 121) but forgot to include a single square in target mask (ep square was included but not the opponent pawn in front of it)
this made it not parse legal enpassant captures out of check
what was impacted: anything related to SAN moves:
- opening pgn games
- reading pgn books (though I'd argue that few books would contain this kind of a move and they'd have to go deep) - worst case a line would be missing or truncated
- moves coming from xboard engines would sometimes be wrongly declared as illegal (=false illegal move claims)
what was not impacted:
- moves coming from uci engines
- saving final tournament pgn (because only a buffer was read, bypassing SAN parser)
after the fix I also verified build 123 against old SAN parser at it found no problems in some old 5 gigabyte lichess pgn database (5m games)
and neither in Ed's million base (3m games), except that the former did contain 3 actual invalid games due to actual illegal moves (that I verified manually)
my apologies again, but bugs do happen sometimes...