They pursued it with forks and hope;
They threatened its life with a railway-share;
They charmed it with smiles and soap."
Maybe this time the elusive cursor hotspot bugs will finally be fixed, once
and for all. The blitCursor() function now calculates both the real and the
virtual dimensions of the cursor, since they may both be different from the
original dimensions.
svn-id: r22913
scaling or aspect-ratio correction, so it has to be added using real surface
coordinates. So I had to re-introduce the extra parameter to addDirtyRect().
svn-id: r22681
each iteration of the loop, causing a serious slowdown on my computer. It
empathically should not be doing that, of course, but to safe-guard against
such patological cases, if the entire screen is dirtied even once we can set
_forceFull to true.
svn-id: r22616
cursor palette in overlay mode. Now it does, which fixes bug #1476666. The
reason it worked at all before was that blitCursor() was called while still not
in overlay mode.
This, of course, means that the GUI can corrupt the in-game cursor palette. I
don't have any game that uses this feature, so I don't know if it's properly
restored again afterwards. But if it isn't, that should be fairly easy to fix.
svn-id: r22539
- Eriktorbjorn's patch from same tracker item for scaling sword1/2 cutscenes
is applied as is. It lacks resolution switch on-the-fly.
- GUI widgets are repositioned on the fly and use most space, even aspect
ratio corrected screen is used without scaling
- Heavy tesing is required, but works for me in all cases except for bug
#1483272: "GUI: SCUMM pause dialog breaks upon scaler switch" which needs more
work.
- I probavly broke some backend or two
svn-id: r22505
loadGfxMode(). As reported on the forum, setting aspect ratio correction
in a 640x480 game would cause ScummVM to crash when the game is started
from the launcher.
With all the graphics transactions and stuff, it's a bit hard to follow
what's going to happen, but I hope deferring this check until the last
moment (i.e. just before the graphics surfaces are actually created) is the
correct bugfix.
svn-id: r19491
except essential Normal1x. Also thumbnails get compiled unconditionally
so we get them on all platforms, even those which never display them.
Sorted out mess brought to sdl backend by epoc maintainer. He misused
__SYMBIAN32__ where DISABLE_SCALERS is more appropriate.
svn-id: r18850
