From 8ffdb8461a98f31a4df8676fd204dad75a750ecd Mon Sep 17 00:00:00 2001 From: Aaron Culliney Date: Thu, 3 Sep 2015 22:23:06 -0700 Subject: [PATCH] Better comment for the current serious deficiency --- src/video/gltouchjoy_kpad.c | 20 +++++++------------- 1 file changed, 7 insertions(+), 13 deletions(-) diff --git a/src/video/gltouchjoy_kpad.c b/src/video/gltouchjoy_kpad.c index 6cf5564d..359617d9 100644 --- a/src/video/gltouchjoy_kpad.c +++ b/src/video/gltouchjoy_kpad.c @@ -116,20 +116,14 @@ static inline void _touch_sourceBegin(volatile int *source) { static void touchkpad_keyboardReadCallback(void) { assert(pthread_self() == cpu_thread_id); - // HACK: filter out a certain number of callbacks to be assured that emulated software is actually ready for a new - // key. - // * J---- appears to work best everytime - // * S---B----- works better every other time (otherwise it misses flip-flop values) - // * DOS3.3 AppleSoft BASIC shell appears to need a much larger value ... - ////static unsigned int callbackCounter = 0; - ////++callbackCounter; - ////if ((callbackCounter % kpad.callbackIgnoreThreshold) != 0) { - //// return; - ////} + // HACK FIXME TODO : // - // Arguably the proper fix HERE to understand the Apple //e keyboard hardware itself and what the maximum rate allowed - // for changing keys. Then this callback would be called by the underlying VM at that rate (by perform the - // appropriate machine cycle counting -- same as for joystick handling) + // There are a number of cases where the emulated software is reading the keyboard state but not using the value. + // This is quite noticeable in a number of games that take keyboard input. + // + // This indicates that we are incorrectly emulating the keyboard hardware. The proper fix for this touch keypad + // joystick will be to properly emulate the original hardware timing, using the existing facility to count 65c02 CPU + // machine cyles. #warning FIXME TODO : implement proper keyboard repeat callback timing if (kpad.lastScancode >= 0) {