|
|
Post by Ryuujin on Jan 31, 2005 2:33:29 GMT 1
If Phoinix can do its own "UDMH" that's safe (and it does), I see no reason why this can't be implemented for LJP. Phoinix is a GB mono emulator for Palm OS 3.5-5. phoinix.sf.netSomeone should ask Bodo Wenzel (the author) how he did it. |
|
|
|
Post by yoyofr on Jan 31, 2005 14:02:42 GMT 1
simple answer: Phoinix is really less complex than LJP, so it has been easier to include such a feature.
technical answer : Phoinix is mono-thread process, since no sound is generated. It's then easy to add a 'umdh' like feature, since the 2 memorysemaphore functions are available for getting write access to feature memory, but will block every other running thread (so no sound).
ljp cannot follow the same way without rewriting a lot
of code in order to "lock" process the shortest possible time. Moreover it will probably lead to noisy sound output.
The correct way is to 'hack' memory functions (allocate, free, copy, ...) so that dynamic heap is redirected to feature memory and write access is given.
|
|
|
|
Post by Tinnus on Jan 31, 2005 17:47:48 GMT 1
Hm... there may be a possibility. From the mailing list...
On Thu, 6 Jan 2005, Dmitry Grinberg wrote:
# MemSemaphore functions while are useful disable sound callbacks for
# streaming sound as well as all other multitasking in the system if
# called with (true) paramater. Better alternatives are
# HwrEnableDataWrites and HwrDisableDataWrites as those peform the same
# function as MemSempahore* functions, but do not disable multitasking,
# allowing sound to continue playing.
#
# PACE docs say that these calls are NOPs in OS5, but that is false, as
# it works!
|
|
