v1.4.7

Stop the engine when displaying a multi-game PGN game list. We used to do this, but it fell out by accident in the great rewrite around 1.4.0.

v1.4.6

Assorted tiny improvements and fixes.

I've no features in the works right now so this may be the last version for a while.

v1.4.5

Clear searchmoves buttons when re-entering a position (except locked position), because it was too easy to overlook them.

v1.4.4

Fix for very rare case of telling the engine to go when we shouldn't (triggered by changing the node limit while the position was a checkmate / stalemate / etc).

v1.4.3

Some fixes in the logic for transitioning between one state (e.g. auto-eval, locked analysis, etc) to another.

I think I'm now finished writing and testing the new logic that had to be added in 1.4.0 to facilitate locking the engine to a position while navigating around. Let's hope.

v1.4.2

• Better behaviour of inferred info when moving forwards from live analysis.
• Be more explicit about what scripts aren't for.

v1.4.1

Nibbler can now run basic "scripts" - which are text files containing raw UCI input to send to the engine. This is intended for Leela devs and other advanced users. It can, for example, be used to set MLH settings with a single click.

It is intended only for engine configuration.

This won't be relevant for most people.

v1.4.0

Relatively minor changes since rc2; most notably, a message is displayed in the infobox explaining when info is inferred. To recap then, the major feature of 1.4.0 is that the engine can be locked to a position, allowing moves to be made without interrupting analysis.

There will probably be a 1.4.1 within a day, which will add a UCI script runner for advanced users.

v1.4.0-rc2

Restore the behaviour we used to have, where the PV and EV of a clicked node in the game tree, if missing, can sometimes be inferred from ancestral nodes and displayed.

v1.4.0-rc1

Allow the user to lock the engine to a position, so that moves can be made on the board without interrupting the analysis. This required a substantial logic rework; please report any bugs.