Now the log() function writes to a file, as
the AGI engine did.
It appears that ScummVM doesn't have any "append" functionality.
As a result, I made this implementation create a new file each time
a message is logged. Note that the file is only created if the
log() function is called. The file name includes the game name
and the current time, ensuring that logs won't be overwritten.
The format is: agi.<gameid>.<currentime>.log and will be logged to
the working directory.
Last week I first implemented the log() function. In that commit
I just had it log to the console. In this commit we start
writing to a file.
This change simplifies the AGI engine volume handling by using the music
and sound fx volumes set in the ScummVM GUI as the maximum levels which
can be set by the scripts. This avoids the complexity of synchronising
changes between these and the scripts (which only had a single 16 step
volume range which is inverted in meaning by quite a large number of
fangames).
This still allows the game scripts to adjust the volumes, but only within
the maximum range set in the Launcher / GMM.
This avoids issues such as https://bugs.scummvm.org/ticket/10732 where
users complained about extremely loud output in some cases.
- These structs are necessary for detection, as well as the engine itself.
- So, we move it to a common header file, include it once in both the files.
- This way, we won't have to include agi.h inside detection translation unit, which saves us much junk code going into the executable.
This fixes these, but adding constructors causes further memset usage
warnings on the structures which are now "non-trivial" due to the addition
of constructors. Should be able to fix by repeating this process to remove
further memset usage.
It was also necessary to make sure that the Common::String objects
were initialised correctly by switching to use a C++ container
for engine objects instead of calloc, since they were no longer
C-compatible PODs.
Thanks waltervn for finding this one.
Was a regression caused by the timer heuristic for detecting
bad script code. When the heuristic identified being in an
inner timer loop, it told ScummVM to sleep + process events.
During that time a restore can get triggered by the user via GMM.
When that happens, the restore is executed immediately.
When still being inside testIfCode(), it may happen that execution
goes beyond the end of the current logic incl. error/crash.
TODO: maybe better change GMM as a whole for AGI, that restores are
always processed in a delayed way after main loop got processed
once?
Original AGI only created a stationary event, when the last
pressed direction key was released. We do that now as well.
Makes cursor in MH1 behave properly.
Thanks to waltervn for pointing this out.
Original AGI TIME_DELAY 0 made the game run as fast
as possible. We didn't go that far, but it still seems to run
too fast. This should hopefully fix that.
See bug #9607
Recently we started to use this as new semantics, although in the past
we used simly <engine>_H. Now these guard defines are consistent with
rest of the files which are used in the engines.
Original AGI Hercules rendering was restricted by the Hercules
resolution, which is why they had to remove the AGI prompt and
replace it with a prompt window like the one used in SCI.
We are not and even when we support that prompt window, we
will still support the original AGI prompt even for Hercules
rendering.
Maybe some games had special graphic code, when Hercules rendering
is active. This needs to get checked.
We are now telling the scripts that we are running on EGA,
so that they don't put text at different coordinates.
Hercules rendering also used a different grid for putting text
for example Space Quest 2 right at the start.
This change avoids problems with such placement. We don't really
want to restrict ourselves by implementing that different
text grid resolution.
Also added comments on game feature flags.
- User option to force Hercules hires font for any rendering mode
- Also change mouse cursor hotspots from 1,1 to 0,0
- Fix inaccuracy in mouse controlled game menu
- Change render_Block(), drawBox(), drawDisplayRect() to use
upper Y instead of lower Y. Original AGI uses lower Y, but
upper Y makes upscaling way easier.
Plenty of fan games set volume to mute, because they thought 15
would be loudest. It's in fact "mute" in AGI. Those games were
made primarily for PC AGI, which did not use the volume setting.
We do, so such games would get muted audio. We try to detect
such games. Hopefully fully fixes bug #7035.
Too many games to try them all out.
Detects when end.of.loop + motions are used on the same screen
object at the same time, which would have resulted in flag
corruption in the original interpreter. We detect this
situation now, show a warning and disable the cycler in case
cycler was activated first.
This solves a new issue in kq1, when grabbing the eagle in room
22, that was previously hidden just like in the original AGI.
Fixes bug #7046
Instead of directly converting play time to VM timer vars,
we only adjust VM timer vars according to play time seconds delta.
This is needed for at least Mixed Up Mother Goose. The game
sets the VM "second" variable to 0 and then uses it for syncing
the text to the songs. We even adjust second calculation, so
that the timing should always be the same and not depend on
basically luck (seconds may wrap right after setting it, because
internal cycles counter is not changed).
And in that case poll events, delay for a few milliseconds and
update screen.
This somewhat worked before the graphics rewrite because of
a timer hack.
This one tries to detect actual inner loops.
Happens in at least Police Quest 1 when playing poker.
Also add a table for Apple IIgs games for music issues like in PQ1
intro, where the developers thought that loading a new room would
take X amount of time and they started loading before the music was
done. In our AGI music may get cut off because of this.
It of course also happens, when using the original interpreter,
as soon as you emulate a faster processor. Which means this
is also a workaround.
See bug #7026
Hold-Key-Mode got introduced v2.425, it was simply not possible
to disable it until 3.098.
Now creating a AGI_KEY_STATIONARY event, so that it works properly
Fixes Mixed Up Mother Goose
Removed pollTimer()
Renamed pause() to wait()
Doing 10 msec delays instead of at least 50 msec per EventProcess
Seems to fix weird Gold Rush ingame timer issue?! bug #4147
processEvents() renamed to processScummVMEvents()
mainCycle() renamed to processAGIEvents()
have.key now sets up an inner loop and calls processAGIEvents()
to avoid any further cycle work processing
This mostly enforces tabs for indentation and spaces for formatting. But also
fixes spaces on empty lines, some extra/missing spaces.
astyle + manual fixup
Daniel Abramowitz
Eugene Sandulenko
Thunderforge
D G Turner
Zvika Haramaty
aryanrawlani28
Henrik "Henke37" Andersson
Paul Gilbert
Bastien Bouclet
Filippos Karapetis
Colin Snover
Martin Kiewitz
Ori Avtalion
Johannes Schickel