Version 1.2
Released: 17.06.2015

Various improvements, bugfixes and cleanups:
- Revisited king safety.
- Revisited extensions and reductions (only check extension in use now).
- Improved time control.

Version 1.1
Released: 10.06.2015

(Changes and notes)

- Corrections to root move ordering scheme.
- Revisited forward pruning.
- Added checks search in quiescence.
- Various bug fixes and cleanups.

Version 1.0
Released: 28.05.2013

(Changes and notes)
- Entirely rewritten engine with magic bitboards.

Version 0.3.1
Released: 24.11.2011

(Changes and notes)
- Added Gaviota tablebases support. For more information, please visit: Gaviota chess engine's website
- Pawn structure evaluation revisited.
- Improved Knight and Bishop outposts evaluation.
- Several code cleanups and optimizations.

(Fixed Bugs)
- Fixed bug in static exchange evaluation, related to discovery checks testing.

Version 0.3
Released: 09.09.2011

(Changes and notes)
- Improved King safety and SEE.
- More aggressive reductions, including reduction for "bad" captures.
- Several code cleanups (like removing the pv array that is not used anymore).

(Fixed Bugs)
- Wrong castling flags update, cause by improperly tested rook movement - fixed.
- Various SEE bugs - fixed.

Version 0.2.2
Released: 21.07.2011

(Fixed Bugs)
- Fixed critical bug storing pv moves and zero values after terminated search.
- King piece-square tables error fixed.
- Fixed critical bug in generate evasions where pawn is tested if on 7th rank instead of 8th after being moved.
- Several other code-cleanups, minor adds and fixes.

- This is a bug fix release and no more than 10 ELO gain is expected!

Version 0.2.1
Released: 01.02.2011

(Changes and notes)

- Root search revisited:
  applied LMR, check extension, Pawn Push extension and basically, the root node now looks almost identically to inner nodes.
- New psq tables for white pawns, black pawns, white endgame pawns and black endgame pawns.
- Several optimization and code cleanups.
- PV displaying - if the principal variation from the triangular array is shorter than the depth passed,
only then the pv is extracted from the transposition table.
- Mate threat extension removed, instead mate_threat is used to restrict LMR and FP now.

(Fixed Bugs)
- Pawn hash table - Preventing the pawn pst value to enter in the table, since it's a dynamic value.
- Mate in # displaying - corrected (STM perspective).

Version 0.2
Released: 01.07.2010

(Changes and notes)

- New time control management scheme. The time distribution formula was inspired by Tord Romstads
from the early versions of Glaurung and achieves two goals incomplete iterations are reduced to the minimum and
time is distributed better between the moves. No other changes. Difference in strength - only in matches.

Version 0.1.9
Released: 16.06.2010

(Fixed Bugs)
- Pawn push extension bug fixed.
- Bug in eval_white_passer fixed.
- Bad captures odering bug in main search fixed.
- Memory Alignment issues for Transposition And Pawn tables were fixed by writing own aligned allocation functions.

(Changes and notes)
- New TT (4 entry buckets) + new replacement policy and better indexing.
- Pawn Evaluation rewritten. The array-fields was replaced by incrementally updated bitboards;
- Added couple of new pawn structure evaluation terms.
- Removed couple of restrictions in IID.
- Several optimizations (Splitting-up the mobility evaluation to omit piece-list loops, as well as several other minors).
- Several code cleanups.

Version 0.1.8
Released: 15.03.2010

(Fixed Bugs)
- Recapture extension removed entirely. Even after the proper implementation
(without the found bug) of the idea, there was no noticeable benefit, so it was removed.

(Changes and notes)
- Small optimization - Dynamic update of the material count, which relieves the evaluation of collecting it.
- Maintenance release.

Version 0.1.7
Released: 02.03.2010

(Fixed Bugs)

- Reset Timer on Ponderhit conception was fixed. False by default.
- Fixed Pondering to do infinite analysis until the stop command is received.

(Changes and notes)
- Evaluation:
    Added center control weights.
  Rook and Knights material imbalance tables.
  New (higher) Knight value.
  Added code for KBNK simple mate.
- Movegen: New legal evasions generator.
- Ordering: KxB and BxK are now considered as equal captures.
- UCI option Ponder added in order to invoke ponderhit in GUIs (like Fritz 11) with no default support of it.
- Better Polling.
- Several code cleanups (like removing the annoying illegal position hack from qsearch()).

Version 0.1.6
Released: 14.02.2010

(Fixed Bugs)

- Wrong calculation of the root moves count fixed.

(Changes and notes)
- Polyglot book support added.
- New attack detection scheme.
- SEE rewritten in accordance with the above.
- Incremental update of the board state (material and pawns).
- Main search and qsearch was revisited in order to eliminate some inaccuracies and errors.
- Transposition table entry is being remade to 16 bytes of size.
- Improved move ordering.

Version 0.1.5
Released: 01.01.2010

(Fixed Bugs)
- Reset Time On Ponderhit command fixed to work properly.
- 1.5 seconds shorter time for the last move in x/x controls to avoid delay and eventual time-loss.

(Changes and notes)
- Reorganized move ordering.
- Dynamic Time Control is now added as UCI option. It is a matter of preference to be turned off and it was designed
to be stoppable mainly for human vs engine games. Turning OFF that option would certainly affect the engines strength (blitz mostly),
but is still not enough tested to be clear how much exactly. The default parameter is true,
because Pawny has always use that mechanism since version 0.1 and the preliminarily test shows that it plays better with DTC ON.
- Developed built-in epd function for the Strategic Test Suite file format in console mode. The commands are visible trough help.

Version 0.1.4
Released: 04.12.2009

(Changes and notes)

- ponderhit UCI commands reception - implemented. The lack of it in the previous versions was costing time losses with ponder ON.
- Added Reset_Timer_On_Ponderhit UCI option, which is OFF by default. When the option is ON and the engine receives
ponderhit command from the GUI, the time for the next move will be allocated again and the search will continue.
- The hashfull commands output is now corrected and displays the amount of saved (or re-written) hash entries during the last search.
- Small search changes.

Version 0.1.3
Released: 12.10.2009

(Changes and notes)

- Board representation changed to 16*16 for better alignment.
- Developed basic pawn hash table with a lot of room for optimization and improvement.
- Pawn evaluation code reorganized to maintain the pawn table.
- Added some trivial draws recognition liable to further improvement.

Version 0.1.2
Released: 01.10.2009

(Fixed Bugs)

- Irrelevant 'info score' with possible value of INFINITY,
after the search is completed costs the game result! -> bug.pgn
- ucinewgame initialize TT again with unknown size fixed.
- Too short history stack fixed.
- Too short uci line buffer, causing an exception fixed.
- Improperly calculated material gain for move in futility pruning fixed.
- History stack improperly indexed in the recapture extension fixed.

(Changes and notes)
- New data structures for move representation.
- Some work was done on the info output, still more to be done.
- A lot of code was revised regarding new types and sizes.
- Better code logic, typecasting and other minor changes.

Version 0.1.1
Released: 15.09.2009

(Changes and notes)

- New move generator (entirely rewritten).
The board is now represented as 16*14 array,
Piece lists - as array of linked lists.
- Evaluation and Knowledge modules was entirely rewritten regarding the
new representation. Same for board initialization functions.
There is NO new chess knowledge added to the evaluation function only code optimizations!
- Several other optimizations have been made, again regarding the 16*14 rep.
(that includes removing the futile outside board checks (during delta attack gen. for example)
and double memory references for getting piece information).

Version 0.1
Released: 03.09.2009

(The Initial version, with the following features)

Board: 0x88 board representation.
- Iterative deepening
- Alpha-Beta PVS + aspiration (1/5 PAWN)
- Internal Iterative Deepening
- Adaptive Null Move Pruning
- Futility Pruning (including Ext. F.P. and Razoring)
- Delta Pruning at Quiescence Search, based on SEE  results (1 PAWN value margin)
- Late Move Reductions (4 Full depth and Reduction of 1 ply)
- Killer and History Heuristics
- Transposition Table (1 table, using always replacement)
- Extensions:
    Check Extension (1 ply)
    Single Reply to Check Extension (1 ply)
    Mate Threat Extension (1/2  ply)
    Queen Prom. Extension (1/2 ply)
    Pawn Push Extension (1/2 ply)
    Recapture Extension (1/2 - pv nodes)
    Fractional Depth Representation (*4)
Material, PSQ tables, doubled, hanging, passed, backward pawns.
Piece Development, Mobility, King Safety, Distance and support for passed pawns.