# Revision history for Mahjong
See git repo for more detailed commit information.
## 0.1.0.0
- First functional prototype.
## 0.1.0.1
- Dependency versions reworked.
## 0.2.0.0
- Help messages improved. Module structure refactored. Documentation begun.
## 0.2.0.1
- Revised cabal version in attempt to fix build error on Hackage. Build works
- locally, suspect the problem is a quirk of Hackage build system.
- Added some more documentation.
## 0.2.0.2
- Internal library now unnamed. Hoping this fixes Hackage build fail.
## 0.2.0.3
- Added docs for yaku functions.
- Reworked seven pairs detection to ensure hand has 14 tiles.
- 9 gates now checks that the hand is closed.
## 0.3.0.0
- Fixed error in scoring triple triplets, was previously scored as 1 han open.
- Implemented new backend for scoring and yaku detection via "context" based
approach, defined in new Context submodule. This helps simplify function
signatures for operations that need many pieces of information besides just the
superficial composition of a hand.
## 0.3.1.0
- Now only ask for dora after we know the hand is not a yakuman, as part of the
mkContext function (previously handled dora in displayHandScore).
- Four concealed triplets now asks about concealment. It wasn't doing this before,
a bug introduced by the refactor, since mkYakumanContext intentionally gets less
information than mkYakuContext since generally it needs less.
- Now only ask for wind context if we see a wind tile in the hand. Note even a
wind pair could matter - it won't affect most yaku but affects Fu.
Speaking of -
NOTE: Seat+Round wind pair counts as a yakuhai pair and awards
2+2=4 Fu. Some rulesets would only award 2 Fu.
## 0.4.0.0
- Extended documentation. Old getFu replaced with new one (previously _getFu).
- Removed old getYaku function
## 0.4.0.1
- Fixed catastrophic failure affecting waits command.
## 0.4.1.0
- Fixed further problems with waits command. Previously, partial hands with kans
were not considered. To fix this and also streamline the internal API, formMelds
now returns all ways of pulling melds out of a hand, INCLUDING KANS. Previously
the caller had to handle kans themself. This makes the interpretHand function
simpler, and fixes getWaits.
## 0.5.0.0
- Sub 5-han mangans are now named correctly
- Reworked meld forming functions. Can now get all ways of pulling out melds, or
just all ways of pulling out a complete set of melds (when the hand is
complete to begin with)
- Internal code refactors according to hlint
## 0.5.1.0
- Form melds now does divide and conquer across different suits.
## 0.6.0.0
- Shanten calculation
- Optimised formMelds by ensuring we don't consider a number of kans that
doesn't make sense with the given hand size. This however does require that
the original size of the hand is known by the function.
- Now have tile efficiency, driven by shanten calculations, which allows for a
"discard" command that recommends the most efficient discards.