| Age | Commit message (Collapse) | Author |
|---|
|
Several functions in Audio32 would call into the mixer to pause
or resume the audio handle, which would cause a deadlock if the
mixer's mixCallback timer fired while one of these functions was
running on the main thread.
To address this, calls to mixer to pause/unpause the digital audio
handle have been removed. Since this was just an optimisation to
prevent unnecessary calls to fill the audio buffer, the only
problem now is that a tiny amount of CPU is wasted on unnecessary
callbacks to read from the empty SCI mixer.
|
|
|
|
Volume was lowered, but not raised afterwards for certain cases
Original script patch was commited in
b676530528349842ad178d02a2cbe49a0facd68e
|
|
Needs lots of testing of course
Applies to 1.51 and 2.00
|
|
Existing saves are compatible but must be renamed to riven-###.rvn
|
|
- Thumbnail
- Save date
- Save description
- Total play time
|
|
|
|
|
|
|
|
The subtitles in KQ7 were not finished and don't quite work right,
drawing off the screen and in weird positions. They could probably
be fixed with some aggressive script patching, but for now, they
are just disabled (as they are in the official release).
|
|
|
|
|
|
Ow. My eyeballs.
|
|
|
|
This fixes digital samples in at least QfG1, SQ3, LSL5. Bug #7159.
|
|
SCI32: Improve FileIO read-write support
|
|
At least Phantasmagoria and PQ:SWAT require this.
This adds a ReadStream/WriteStream to support this. It replaces
the previous VirtualIndexFile which required reimplementation of
most read/write functions.
|
|
Wasn't bug #6571, but bug #6751
Thanks to omer_mor for spotting it
|
|
This reverts commit 61a8a5a88f4f02f1d4cdd3523ad6cec63701d8dd.
|
|
|
|
|
|
Now the Riven options dialog has the same features as the Myst one:
- Screen centering
- Save / Load / Quit buttons
|
|
|
|
|
|
Fixes bug #6571
|
|
|
|
|
|
The sound is rather bad, but I assume that's something with our
CinePak video decoder so I'll leave it like this for now. The player
itself is based on the BBVS engine.
|
|
This avoids a conflict between the "all" directory and the "all"
archive, if the directory happens to be named in all lower case.
While it isn't on the CD, we shouldn't introduce case sensitivity
in the file names if we can avoid it.
|
|
At least in my copy, the voices directory is on the same level as
the other directories. (Though the whole thing is inside a
directory called "data", so maybe that's where that came from?)
|
|
SCI: Clean up some aspects of call handling
|
|
MISC: Bugfixes
|
|
Controls that manage their own event loops and call frameOut
directly generally need to sleep in order to avoid 100% CPU,
just like the main VM event loop.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The ExecStack constructor set argp[0] to argc before. This is now moved
to the caller, to make this action more explicit.
|
|
Modifying a value above the stack pointer doesn't seem to make much
sense. This was added in FreeSCI back in 2002 in a pair of commits that
did not make clear what the purpose of this was. My guess is that it
attempted to adjust argc, but failed. This wouldn't have been noticed
since argc was always set correctly by make_exec_stack_entry (which is
now the ExecStack constructor).
|
|
ExecStack now stores the kernel call number as well as the subfunction.
This allows kStub and backtraces to log the actual subfunction called.
The kernel call number in ExecStack used to be stored in the
debugSelector field. It now has its own field, to avoid confusion.
|
|
|
|
|
|
|
|
The way dimensions of scaled screen items are calculated changed
over the lifetime of SSCI. In early low-resolution and
mixed-resolution games, scaled drawing needed to use at a global
cadence across the entire screen to ensure proper alignment, but
in later games (like Torin), local scaling of individual screen
items seems to be the way scaling is performed.
|
|
Using the one from SCI2.1mid makes fades very slow because SDL has
a larger audio buffer than SSCI DOS. This new algorithm is based on
wall time so will always fade at the correct speed, although the
larger buffers will have a coarser granularity so the fades may
not be as smooth as in the original engine. If anyone cares, the
fade volume could be mixed into individual samples in `readBuffer`
instead of applying just once per complete buffer. SSCI did not
do this, however, so this implementation should be pretty accurate.
|
|
|
|
|
|
|
