They're generally the largest resources in the cache by far (though some
ANIMATION_FILE resources are about as big).
I still don't know how much benefit there is to resource caching, but some
of it is definitely needed, or the game won't work properly. Oh well, as
long as no one complains about the extra memory usage...
svn-id: r15079
command, would close the global script variables and player object
resources, without reopening them again. This made them fair game for the
resource expiration mechanism. The player object is probably referenced
often enough to stay alive, but the variables died on me pretty quickly,
causing ScummVM to crash.
I've also added a "reslist" debug command to make this sort of things
easier to spot. By default it only lists resources with refCount > 0. Use
"reslist 0" to see all the cached resources as well.
svn-id: r14958
new memory manager didn't have the concept of the NULL pointer. Now it
does.
If ScummVM ever crashed for you when using the phone early in the game,
this patch hopefully fixes that bug. (If it didn't crash for you, memory
block zero was still allocated, so 0 still decoded to a valid pointer.)
svn-id: r14937
to fix, but it should work well enough for now.
In this rewrite of the music code, I removed the "save/restore music state"
function, since it just complicated things for a very small gain. It wasn't
in the original engine, and I added it just for the credits, so that the
previously playing music could be resumed afterwards. I might re-add it
later, but probably not.
svn-id: r14887
class, so they are handled the same way as the compressed clusters.
The next step will be to migrate the music playback to use the same class,
which means the fade-in/out logic needs to be separated from the decoding.
Once this is done, adding support for compressed music should be a piece of
cake.
svn-id: r14740
sprites on exit. As far as I can tell, the only case when this makes any
difference is when there is text on screen when you quit ScummVM, so it's
not really a memory leak, but Valgrind will report it as one.
svn-id: r14738
speech. (Apparently it was just an accident that MP3 worked.)
Unfortunately I had to change the file format of the compressed files to
include both the compressed and uncompressed size, but since the tool to
create these files has only lived as an item in the patch tracker, no one
should have exptected it to be the final, working version, right? Right.
svn-id: r14698
this once. :-)
The parameters to drawLine() aren't clipped to the screen size, which meant
that it was accessing memory out of bounds when marking the screen as
dirty. The function now uses plotPoint(), which does the bounds checking
and screen dirtying for us. Apart from being a little easier to read, it
dirties only the parts of the screen the line actually passes through,
instead of a rectangle defined by the line's end points.
Since drawLine() is only used for debugging, I wouldn't consider this a
particularly serious bug.
Next change is that only the pixels inside the original parallax layer are
considered when creating the block surfaces. This may make the drawing
slightly more efficient, since fewer surfaces will be labelled as
transparent.
Plus some other minor cleanups.
svn-id: r14340
by accident, and could cause bad noises during music cross-fades.
This wasn't a problem in 0.6.0 since all music is sampled at 22050 Hz,
which is the most likely output sample rate for ScummVM, so the converter
didn't actually have to do anything. Now, however, the output sample rate
could be anything.
I've given the music streams one converter each. In BS1, which uses similar
music code, it was already necessary to do this since some of its music is
sampled at 11025 Hz.
svn-id: r14237
game to crash shortly after Andr� shows you the coyote stone. More
precisely, when the camera view shifts from the close-up of the
conversation back to the normal view of the caf�.
For those who enjoy reading commit messages, this was the crash I was
hunting for yesterday.
svn-id: r13954
or presses a button. This is how displayMsg() was always used, so the only
difference is that the code to check for events is no longer outside the
function.
In the process, it turned out that removeMsg() was probably unnecessary so
I have removed it. May cause regressions, but we can deal with them later.
svn-id: r13953
found the old name misleading (there is only one array that stores the
palette in the engine, though it could be argued that it's a copy of the
one used by the backend), and removed some code that I'm almost certain was
never used. (I've added assert()s to trigger in the cases where it would
have been used.)
svn-id: r13949
our other engines do this, so there is little reason for BS2 to. I did add
a filtering mechanism so that mouse button releases and scroll wheeling is
ignored during normal gameplay, but I don't know if that was necessary
either.
Since this left little more than an empty husk where the Input class used
to be, I've eliminated that class and buried its remains in Sword2Engine.
svn-id: r13812
so that it gets properly redrawn. Only the debugging code uses these
drawing primitives, so it's no big deal, but it's still the right thing to
do.
svn-id: r13811
didn't do the change I was hoping for: the coyote stone is still partially
see-through, but perhaps it was in the original as well.
At least we no longer need to keep the buffer the mouse cursor is decoded
to, since that's now handled by the backend.
svn-id: r13782
Part of this cleanup involved removing _unpauseZone. It was only used by
fnISpeak(), and as far as I could tell it was just because the original
code didn't trust amISpeaking() and getSpeechStatus() to return sensible
values directly after unpausing the game.
svn-id: r13781
to keep its own copy of the sound data. It could be even further simplified
(I don't really see any reason for having two different sound queues), but
I seem to have reached a point of stability here and I don't want to jinx
it by making further changes yet.
svn-id: r13705
