Age | Commit message (Collapse) | Author |
|
svn-id: r8780
|
|
svn-id: r8779
|
|
svn-id: r8778
|
|
now in util.*)
svn-id: r8777
|
|
svn-id: r8771
|
|
svn-id: r8764
|
|
svn-id: r8753
|
|
conditions
svn-id: r8740
|
|
to invert their meanings. This is useful for overriding settings in the
config file.
svn-id: r8733
|
|
svn-id: r8727
|
|
particuarly bad if timers are implemented via threads), plus this should help in fixing race conditions in classes using class Timer
svn-id: r8722
|
|
the number that was requested. A lot of our code obviously assumes this
behaviour, and the MP3 file player depends on it because otherwise, when
reaching the end of the file, it will think it read a full buffer, when in
fact it only got part of it and the rest is the remains from the previous
read.
This would cause a brief note from earlier in the track at the end of the
MI1 chapter one screen. Probably other cases as well.
Of course, this may cause regressions but only - I think - if the code was
buggy to begin with.
svn-id: r8721
|
|
one (good quality with acceptable speed)
svn-id: r8715
|
|
svn-id: r8667
|
|
svn-id: r8666
|
|
svn-id: r8628
|
|
correct - I was wrong :-) just plug in 0xFFFFFFFF to see it (roughly spoken, we can loose the uppermost bits if we shift after the addition instead of before it)
svn-id: r8627
|
|
svn-id: r8623
|
|
correction mode by making sure that dirty rects always start on a line
that won't have its colours changed by the stretching function. This was
probably only noticeable when using the 1x scaler.
[*] Actually, the glitches weren't specific to the mouse cursor, but it was
useful for triggering them.
svn-id: r8621
|
|
svn-id: r8620
|
|
cursor in aspect-ratio correction mode.
svn-id: r8618
|
|
svn-id: r8616
|
|
svn-id: r8613
|
|
for opengl backend
Enable poll_event in opengl backend
svn-id: r8612
|
|
operationally (read: taking into account side effects and special cases in computer arithmetics) equivalent to the old code
svn-id: r8606
|
|
svn-id: r8605
|
|
SuperEagle
svn-id: r8604
|
|
SEEK_SET mode for conveniance
svn-id: r8587
|
|
svn-id: r8557
|
|
Fix properly propogates previously encountered value of multi_midi
and native_mt32 so follow-on config domains that don't include
multi_midi or native_mt32 won't reset the settings.
svn-id: r8545
|
|
svn-id: r8526
|
|
available.
svn-id: r8521
|
|
svn-id: r8512
|
|
specific stuff from gameDetector.h to scumm.h; cleaned up some #includes (ideally, never #include something in a header file if you can avoid it - this cuts down interdepencies)
svn-id: r8496
|
|
svn-id: r8488
|
|
svn-id: r8487
|
|
augmented by some more changes of mine
svn-id: r8482
|
|
casting it to something non-const is bad, and undermines the compilers aliasing detection
svn-id: r8481
|
|
svn-id: r8480
|
|
under win98
svn-id: r8478
|
|
svn-id: r8459
|
|
svn-id: r8454
|
|
Now all transient config domains use a name
in the form _TRANSIENT_DOMAIN
svn-id: r8452
|
|
svn-id: r8447
|
|
Again, I hope I'm manipulating the domain-based config architecture
correctly. I've created yet another transient domain called
"_COMMAND_LINE", so that after game-specific options are loaded,
the command line options can be loaded and override the config
file. GUI options are then loaded so as to override command line
options.
svn-id: r8437
|
|
Implement true music driver selection from the GUI options dialog.
This implementation depends on a new, transient config domain called
"user-overrides". Probably any other config changes made from the
options dialog will need to go into this domain, which is now
loaded after the game settings are loaded.
I hope this is the right way to do it. I'm still a little fuzzy
about our game settings architecture. If anybody suddenly gets
a "[user-overrides]" section showing up in their scummvm.ini or
scummvm.rc files, you'll know why.
svn-id: r8434
|
|
svn-id: r8431
|
|
svn-id: r8429
|
|
be modified, but can't help it)
svn-id: r8409
|
|
svn-id: r8398
|