Hi David,

You wrote to Leslie Rhorer:

DN>I've received the scoop from IBM and have the basic semaphore
DN>interface working on my test-bed programs here. I wasn't able to use
DN>QBASIC very readily, since CALL ABSOLUTE stinks. As a result my test
DN>code is in C, compiled using Borland 3.1 and Watcom 10.6. The Borland
DN>compiler often generates some bad object code when manipulating the
DN>32-bit semaphore handles, so I am currently sticking with the 16-bit
DN>Watcom compiler for testing.

Borland 3.1 generates duff code if set to allow 386 code generation (it
can't do arithmetic on longs correctly :-( ). This took me some time to
track down in a large program when I added some code using long
variables (in fact they were file seeks and writes, it was seeking out
to 50 Meg in what was supposed to be a 10 Meg file, when it was actually
supposed to be accessing inside the first 100 K or so). If the code
generation is set back to 286 everything works OK (well, I've not yet
found any bugs with my C programs that I can blame on the compiler
rather than my code).

George

 * SLMR 2.1a * Desk: A very large wastebasket with drawers.

--- Maximus/2 3.01