mame_machine_manager::instance()->ui()
Dynamic casts are never one's first choice, but this is superior to
grabbing a global singleton. Upon merging more worker_ui changes, it
will be possible for machine().ui() to be an implementation of
ui_manager distinct from mame_ui_manager, so this code is being changed
to gracefully handle this scenario. Plus, global singletons like
mame_machine_manager::instance are just plain ugly.
Lastly, I would like someone familiar with the LUA integration to look
at this, not just for correctness, but to validate my approach of
keeping LUA's mame_machine_manager::ui() equivalent returning the
mame_ui_manager while changing how it is implemented. Even if getting
rid of mame_machine_manager::ui() is the right way to go, there is an
argument to be made that we should change the LUA integration even if it
breaks compatibility with existing scripts (though obviously I did not
take that approach).
* Changed the various usages of UI_*COLOR to be calls to
src/frontend/mame/ui/moptions.h
The various UI_*COLOR macros were implemented as calls to
decode_ui_color, which cached the values for the various options in a
static array, which was obviously a gross hack.
This refactoring is strategic because I am trying to confine awareness
of mame_ui_manager to code in src/frontend/mame/ui, and the
implementation of decode_ui_color() relied on the ability to access
mame_ui_manager as a singleton from outside this code.
* Created a ui_colors object, so that queries for UI RGB values would not
always require parsing strings
* Replaced UI_TARGET_FONT_[ROWS|HEIGHT] and UI_BOX_[LR|TD]_BORDER macros with
property calls
These macros were implemented with a call to a function
(get_font_rows()) that opportunistically stashed the results of option
accesses in static variables; in other words, a gross hack
Because get_font_rows() attempted to access mame_ui_manager as a
singleton, it was an obstactle to providing an alternative
implementation of ui_manager
* Remove stray debugging cruft that found a way into the other PR
---------------------------------------
ibm5150: SimEarth: The Living Planet (Europe, 3.5", v1.0), SimEarth: The Living Planet (5.25", v1.1), SimEarth: The Living Planet (3.5", v1.1), SimEarth: The Living Planet (3.5", v1.3) [Justin Kerk]
odyssey2: AMOK!, Calculator!, Pong for Videopac and Odyssey²! [Justin Kerk]
- Added new items from the redump.org database:
Cat's Part-1
DOR Best Selection Gekan
Gokko Vol. 1 - Doctor
Professional Mahjong Goku
Scavenger 4 (1993-11-11)
Taiken Shiyou! Marty Channel
- Replaced the following items with images that match the redump.org
database. In many cases the existing images had no offset correction
(which leads to small chunks of audio data at the start/end of the disc
being missing) and no proper pregaps, so these should be better.
The 4th Unit 5 - D-Again
The 4th Unit 7 - Wyatt
Genocide Square
Gunship - The Helicopter Simulation
The Legend of Kyrandia (promoted to working due to this change)
Kyrandia II - The Hand of Fate
Last Armageddon CD Special
Mahjong de Pon!
Misty - Meitantei Toujou
Puyo Puyo
Schwarzschild IV - The Cradle End
Silent Möbius
Splatterhouse
Ultima Underworld II - Labyrinth of Worlds
Wing Commander
Wizardry V - Heart of the Maelstrom
- Added alt_titles and release dates for almost everything
- Miscellaneous metadata fixes
* New working machine added
----------
Game & Watch: Gold Cliff [algestam, hydef, Ryan Holtz]
* hh_sm510: Remove ambigious comment from gnw_gcliff (nw)
* Changed the various usages of UI_*COLOR to be calls to
src/frontend/mame/ui/moptions.h
The various UI_*COLOR macros were implemented as calls to
decode_ui_color, which cached the values for the various options in a
static array, which was obviously a gross hack.
This refactoring is strategic because I am trying to confine awareness
of mame_ui_manager to code in src/frontend/mame/ui, and the
implementation of decode_ui_color() relied on the ability to access
mame_ui_manager as a singleton from outside this code.
* Created a ui_colors object, so that queries for UI RGB values would not
always require parsing strings
* Converted a few more options().zyx_color() to colors().zyx_color()
* A few more misses from earlier
* spectrum_cass.xml - various changes [see credit list]
added some more images to the 'unsorted' section, moved some images to 'sorted' section, identyfing and replacing several bad dumps along the way
Credit for this stuff should be
Steve Brown, Andrew Barker, Paul Hurd, Tony Barnett, Claus Jahn, Donald Graham,
Guy Bearman, Jose Luis Soler, Krasimir Hristov, Andrew Blood, Dave Stevens,
John Wilson, Edwina Brown, Walter Pooley, Tomaz Kac, Martijn van der Heide,
Ramsoft, Dunny, Woody, Mikie and many many others.
* restore clone relationship (nw)
* fix validation (nw)
* details (nw)
* typo (nw)
* filling in some details (nw)
* filling in some details (nw)
* another compilation (nw)
* move more (nw)
* compilation details (nw)
* typo (nw)
* mention set name (nw)
* more (nw)
* move movement (nw)
* a few more (nw)
* more (nw)
* move more (nw)
* more (nw)
* move more (nw)
* more (nw)
* more (nw)
* a few more (nw)
* (nw)
npwoods