> On Aug 25 06:15 94, David Nugent of 3:632/348{at}fidonet wrote:

 > DN> made a blocking call (say KbdCharIn with IO_WAIT), then
 > DN> almost all other Kbd*() calls will block until you press a
 > DN> key.

 > Yes. IBM changed this in OS/2 2.10!

Indeed. :-/  But at least now this agrees with the documentation.

 > Earlier there were NO
 > problem to read the keyboard in one thread, while another
 > thread asks if eg. NumLock is currently pressed.
 > But still working is low-level Device IO Code. So now, i'm using

 >     cRecords = KBD_READ_WAIT | 1;
 >     DosDevIOCtl(&kbci, &cRecords, KBD_READCHAR,
IOCTL_KEYBOARD, _hfKbd);

 > in my keyboard read thread, which enables other threads to
 > use KbdChar* calls.

Ah, that's nice - didn't think of that approach. Instead I just turned on
shift state reporting and let KbdCharIn() do it's stuff, which includes
updating the shift state changes.

david

---