The implementation was not correctly rounding the scaled position
with mulru, leading to occasionally incorrect hit detection at
the boundaries of boxes.
This adds a slightly more accurate comparison algorithm that will
at least ensure that all the engine-generated planes and screen
items with matching priorities will be sorted above
script-generated planes and screen items, like in the original
engine. It still does not sort script-generated items by memory
handle order, so if that is ever a thing that actually happens,
those may still be in the wrong order.
Other platforms, that support analog stick + touch screen at the
same time, are possibly also affected.
Cursor workarounds exist for qfg1vga, qfg3, lsl5 and Island of
Dr. Brain. Those sometimes worked and sometimes didn't on
at least OpenPandora and should be fixed now.
The original engine had some members on PlaneShowStyle to
allow wall clock timing but never actually used them in the
processing loop so transitions simply ran as quickly as the
CPU could process them. For the moment, we will just limit
these transitions to ~30fps, which hopefully roughly matches
the speed of the engine on hardware of the era.
Adding consts to try to isolate rectangles that change during
draw list processing, to be investigated for incorrect mutations
causing lower screen items to be drawn over higher screen items at
their edges.
This implementation is not 100% engine accurate, but it is
more accurate than what was there, and hopefully the differences
between this and the engine code are merely cosmetic.
The known (intentional) differences are:
1. Uses ScummVM rects inside the engine code, converting to/from
SCI rects on the kernel edges and when scaling
2. Fewer side effects when performing operations that *should*
have been pure from the start (like text dimension calculation).
Still not side-effect-free, but at least things like colours
and alignment do not need to be reset every time a measurement
is taken, unlike in the actual engine.
Editor controls and some other kBitmap code are temporarily
disabled as a result of changes to GfxText32 until they can be
updated to be engine-accurate.
- Add "kAnimate fast cast state" to "version" debug command
- Make it possible for script patcher signatures to get fully
used outside of the regular script patcher
- Remove previous fastcast detections and replace them with
a signature heuristic
- Remove object name checking, when fastcast global is set
- Heuristic detects "fast cast" support incorrectly for multilingual
KQ5, but it seems the game never sets the global, so it won't
matter. KQ5 CD (also SCI1 late) has fastcast support.
- Remove hack in GfxView::draw
- Add lots of comments to ScriptPatcher class
This fixes EcoQuest 1 Floppy showing the anemone on top of the
message box (see bug #5170)
- Patch game super object for saving instead of game object
- Remove re-adding planes+screen items, game::replay does that
We would only have to do that for soft-failing on restore
- Change debug levels of kNumLoops/kNumCels to 9
- Add special comment about -info- selector in syncWithScripts()
This should now make ScummVM menu saving work properly at
least in SQ6.
Fixes restoring saved games, when ScummVM dialogs are patched in
Removing clearing planes+screen items as well, because the scripts
actually did that, but did not clear everything.
Fixes "Invalid screen item" message when entering options menu,
after restoring in SQ6.
Colin Snover
Johannes Schickel
Filippos Karapetis
Willem Jan Palenstijn
Martin Kiewitz