[comp.sys.m68k.pc] C compiler floating point

mknox@NGP.UTEXAS.EDU (Margaret H. Knox) (07/13/87)

Larry,

  [This mailer will not send to MOST of the people who send to me.]
  I don't know who you talked to, but the CP/M-68K ALCYON  compiler
has supported floating point since version 1.2.  It is true that the
original version 1.1 did NOT, but then it had so many bugs that it\
was practically unusable anyway (never DID get GETCHAR to work right
in that version).  Version 1.2 had a great number of bugs fixed, and
added both IEEE and Motorola format floating point.  Version 1.2 of the
system also added type-ahead and a number of other features.  A VERY
BIG IMPROVEMENT.

  Version 1.3 was a minor maintenance release, no new features.
And, curiously enough, they BROKE the floating point in the C
compiler!  We are still using version 1.2 at TriSoft, since we can
not get DRI to look at the 1.3 floating point problem.  They and
Alcyon are now concentrating on the ST.

  As far as porting the C compiler (and ST products in general) over
to a standard CP/M-68K system, it could be done.  It would not quite
be trivial, however.  To go either way requires a shell to map the
system calls back and forth (different TRAP numbers, and arguements
passed on the stack for the ST, registers for the CP/M-68K).  But it
could be done.  Compilers and the like are more portable than applications
because they do only fairly simple I/O.  No requests for the disk
allocation vectors and the like.  Also, most of them do not do any fancy
screen graphics calls.

  If you are indeed running version 1.1, you might give serious consideration
to upgrading to 1.2.  Especially if you use the C compiler a lot.  And
you could also investigate the other C compilers for CP/M-68K (Reams,
Whitesmith, etc.).

						jmk

.w