mirror of https://github.com/nmlgc/ReC98.git
98 lines
2.5 KiB
C
98 lines
2.5 KiB
C
/* ReC98 Research
|
|
* --------------
|
|
* PC-98 keyboard key hold behavior test
|
|
*/
|
|
|
|
#pragma option -ms
|
|
|
|
#include "master.hpp"
|
|
|
|
#if defined(__TURBOC__) && defined(__MSDOS__)
|
|
// Remove C++ exception handler bloat on Borland compilers
|
|
// (see https://community.embarcadero.com/article/technical-articles/162/14700)
|
|
void _ExceptInit(void)
|
|
{
|
|
}
|
|
#endif
|
|
|
|
// Should cover the number of possible input states within a frame
|
|
// ⌈(19200 baud / 11 bit) / 56.4 fps⌉ ≈ 31, but let's rather be extra safe...
|
|
#define STATES_MAX 64
|
|
|
|
void hr(int len)
|
|
{
|
|
int i;
|
|
dos_puts("\r\n");
|
|
for(i = 0; i < len; i++) {
|
|
dos_puts("\x86\x44");
|
|
}
|
|
dos_puts("\r\n");
|
|
}
|
|
|
|
void frame(void)
|
|
{
|
|
char states[STATES_MAX];
|
|
char state_prev = 0;
|
|
unsigned int states_recorded = 0;
|
|
unsigned int i = 0;
|
|
int flipped = 0;
|
|
|
|
vsync_Count1 = 0;
|
|
do {
|
|
int delayloop = 1024;
|
|
unsigned char input = *(volatile unsigned char far*)(0x531);
|
|
states[states_recorded++] = (input >> 2) & 0xF;
|
|
|
|
while(delayloop && vsync_Count1 < 1) {
|
|
__asm out 0x5F, AL;
|
|
delayloop--;
|
|
}
|
|
} while(vsync_Count1 < 1);
|
|
|
|
dos_puts("\r \r");
|
|
for(i = 0; i < states_recorded; i++) {
|
|
char state = states[i];
|
|
|
|
if(state_prev != state && i > 0) {
|
|
flipped++;
|
|
dos_puts("\x1B[17m");
|
|
} else {
|
|
dos_puts("\x1B[0m");
|
|
}
|
|
|
|
dos_putch((state < 10) ? ('0' + state) : ('A' - 10 + state));
|
|
dos_putch(' ');
|
|
state_prev = state;
|
|
}
|
|
if(flipped >= 2) {
|
|
dos_puts("\r\n");
|
|
}
|
|
}
|
|
|
|
void __cdecl main(void)
|
|
{
|
|
vsync_start();
|
|
hr(79);
|
|
dos_puts(
|
|
"Hold any of the cursor keys. The line below will show all states of the cursor\r\n"
|
|
"keys, read using the keyboard state memory bitmap at 0x52A (which is also how\r\n"
|
|
"TH02 and later read input), within a single frame.\r\n"
|
|
"The 0.6ms delay between the states roughly matches the keyboard UART delay.\r\n"
|
|
"\r\n"
|
|
"After the typematic delay, you might see the occasional wrong state popping up.\r\n"
|
|
"This will be highlighted in\x1B[17m red \x1B[0mand kept on screen when it happens.\r\n"
|
|
"Unlike on IBM PC/XT/AT where holding a key repeats only the down scancode,\r\r\n"
|
|
"PC-98 keyboards repeat up AND down scancodes, leading to these state flips.\r\n"
|
|
"So, polling the keyboard twice within a frame, ~0.6ms apart, is indeed a good\r\n"
|
|
"idea for a game engine, which is why ZUN chose to do it from TH03 onwards.\r\n"
|
|
"\r\n"
|
|
"Hold Q to quit."
|
|
);
|
|
hr(79);
|
|
while(!(*(volatile char far*)(0x52C) & 0x01)) {
|
|
frame();
|
|
}
|
|
hr(79);
|
|
vsync_end();
|
|
}
|