Near the end of the game, RAMA will start trying to store some
invalid references. This does not affect the save game negatively
in any way, but it was causing the kernel to assert a signature
failure.
Despite what game script disassembly lead me to believe, the game
seems to create only one auto-save, which ends up being saved as
911.sg (not autorama.sg). This save file is created just before
entering the underground Avian Lair in New York, and seems to be
designed as some emergency backup since entering the Avian Lair is
a one-way trip.
Which were originally only played, when using the Windows interpreter.
Afaik the DOS interpreter was incapable of playing more than 1
sample at a time, which was probably the reason why Sierra only
added these for the Windows interpreter.
For example a sample is now played, when points are awarded.
Those samples are currently used all the time and will also
enhance the game, when platform DOS is chosen.
In case someone wishes to opt-out of this, we could add a game
specific option.
We do something like this for Space Quest 3 already,
that's why I don't see a reason to add a game option right now.
In a couple of places, Lighthouse updates the renderer with screen
items for the next room before the room transition video plays.
This is normally fine when using the compositing video renderer
because the videos are drawn into new planes which occlude the
screen items, so the screen items are culled from the draw list
and do not submit their palettes. However, when in HQ video mode,
we currently force the overlay renderer, which was not blocking
screen items before forcing a frameOut, so the screen items'
palettes got submitted prematurely in this case and caused bad
rendering after the video finished playback.
Now, if we are forcing into the overlay code path, we still create
a blank plane behind the overlay before the forced frameOut in
order to correctly occlude screen items and keep them from
participating in rendering before they normally would.
Fixes Trac#10233, Trac#10235.
Previously, notifying bots to the end of a speech fragment was
done in ~TTtalker. Which caused problems when in progress talkers
were freed when exiting the game with a speech was in progress,
since it would try to start the next following speech fragment.
Somehow, up until trying to view an encrypted data cube in RAMA,
the missing relocation of locals did not seem to cause any trouble
in any of the other SCI3 games.
This reverts commit d2dd942ad7.
The Serializable class is an interface class and has only the
empty default copy constructor. There is no reason to call this
explicitly. Whatever is warning about this should be fixed to stop
warning about it, rather than adding such unnecessary calls to the
engine code.
kArrayCopy would perform an unnecessary memory copy of the source
array, the treatment of the count value as unsigned was clearly
not correct since it was being sign-extended and checked against
-1.
RAMA has its own custom save game format that game scripts write,
but we still want to be able to use these save game files from the
ScummVM launcher, so the metadata has to be able to be written
separately from the rest of the game saving.
The maxSize given when generating a SegmentRef is supposed to be
the maximum size of raw data that can be held inside the offset-
part of a reg_t, not the entire size including the segment.
This fixes a buffer overflow in "Inside the Chest", which still
does not work, but at least doesn't cause heap overflows anymore
with this change.
Colin Snover
Martin Kiewitz
Paul Gilbert
Bastien Bouclet
Thierry Crozat