It is used (only?) in Beat Blades Haruka, when two enemies are found by
enemy seek.
This is just a screen effect and can be ignored without affecting
gameplay.
Some games (e.g. Toushin Toshi 3) malfunction when system.GetTime()
returns a negative number.
On many systems clock_gettime(CLOCK_MONOTONIC) returns the time since
OS startup, but this wraps to a negative number after about 24.8 days.
As a mitigation, use SDL_GetTicks() instead, which returns the elapsed
time since game startup.
Since this is the surface to render to, it should not be added to the
list of surfaces to draw from.
This is done by marking sprite -1 as hidden. Now SACT2.SP_GetShow(-1)
returns 0 and SACT2.SP_SetShow(-1, true) raises a runtime error. This is
consistent with the AliceSoft's implementation.
On Android, when the user enters "ら", "ん", "す" with the virtual
keyboard and converts it to "ランス", xsystem4 receives the following
events:
SDL_TEXTINPUT("ら")
SDL_TEXTINPUT("ん")
SDL_TEXTINPUT("す")
SDL_KEYDOWN(backspace)
SDL_KEYUP(backspace)
SDL_KEYDOWN(backspace)
SDL_KEYUP(backspace)
SDL_KEYDOWN(backspace)
SDL_KEYUP(backspace)
SDL_TEXTINPUT("ランス")
The backspace key events are generated by SDL to delete the unconverted
text. However, System4 games process the backspace key by polling, so
if the SDL_KEYDOWN and SDL_KEYUP events occur in close succession, the
game cannot detect that the backspace key has been pressed.
To solve this problem, this patch queues these consecutive input events
and processes them with an interval of at least 10ms.
Fixes https://github.com/kichikuou/xsystem4-android/issues/2.
This adds gestures that allow touch device users to emulate the
following mouse/keyboard actions:
- Right-click: touch outside the game's viewport
- Ctrl key: one-finger touch and hold (for one second)
- Mouse wheel: Swipe up / down with two fingers
A typical event loop in System4 game looks like this:
while (true) {
SACT2.Mouse_GetPos(x, y);
// Update the screen according to the mouse position (x, y)
...
SACT2.Update();
if (SACT2.Key_IsDown(VK_LBUTTON)) {
// A button was pressed at (x, y)
...
}
}
This works poorly with touch devices. Assume that a touch event occured
during SACT2.Update(). It updates internal mouse position and button
status. After SACT2.Update(), the game calls SACT2.Key_IsDown(VK_LBUTTON).
It returns true, but the game has not yet called SACT2.Mouse_GetPos()
after the touch event, so it assumes that the button was pressed where
the mouse pointer was before the touch.
In order to avoid this situation, this defers mouse button events
synthesized (by SDL) from touch. Such deferred events are processed when
SACT2.Mouse_GetPos() is called or after 50ms.
Writing save files in RSM format is only enabled if `--save-format=rsm`
command line flag is specified, because this implementation does not
create save files that are fully compatible with the AliceSoft's
implementation:
* There is no simple way to determine the RSM format version used by the
game.
* There are several unknown fields in the RSM format.
* Empty arrays are represented as NULL pages in xsystem4, so type
information of such objects cannot be saved.
These limitations are not a problem for the following use cases:
* Use the created save files only in xsystem4
* Load save files created by System40.exe
Also, RSM save files are much smaller and faster to read and write than
JSON.
BanMisc.{Save,Load}Struct now serializes structs in the same format as
AliceSoft's implementation, rather than in JSON format. The old JSON
format can also be loaded for backwards compatibility.
For some reason, this function is renamed from GetHDDVolumeNumber only
in Momoiro Guardian.
This also makes all Confirm3 functions HLL_QUIET_UNIMPLEMENTED.
* The "type" argument of S_MOD is 48 for bool, 56 for long int
* %b, %c, %d can consume any integer types (int, bool, or long int)
* If the value cannot be consumed, S_MOD should return the format
string instead of an empty string
It is implemented as a post-processing effect. The hack for the
projection transform matrix has been removed, because now the image is
flipped upside down in the post-processing stage.
Some games (Ojou-sama o Iinari ni Suru Game and Double Sensei Life, as
far as I know) pass "SaveData\\MsgSkip.asd" to MsgSkip.Init. So this
change:
* uses savedir_path() to resolve the relative path, and
* removes the "SaveData\\" prefix, otherwise "SaveData/" will be
duplicated in the resolved path.
This also rewrites the SetWindowSetting and GetWindowSetting functions
based on the understanding gained from examining games after Haru Urare:
* Setting IDs are consistent across games.
* Settings 5 and 6 are added in Pastel Chime 3, both default true
kichikuou