Date   : Mon, 19 Dec 1983 21:53:43 EST (Mon)
From   : Mike Ciaraldi <ciaraldi@Rochester.ARPA>
Subject: DRI C Compiler Update

Here is the latest on the Digital Research C Compiler for CP/M-86.
We have been using the 2nd release of the complier since October.
It works pretty well, except for the floating point!
Floating-point I/O is all messed up. You try to read a
number and get back a zero, for example.
Printing is uncertain, and so is integer<->float conversion.
The Taylor people wound up writing conversion and I/O routines
to work temporarily.

A letter arrived last week saying that version 1.1 would be shipped
in Mid-December. This fixes all the bugs, they say.
It was also used to implement the C and Fortran compilers, so is
supposedly well tested.

This release is the first that uses the front-end/back-end concept.
The Fortran and C compilers share the same code generator (back-end)
but have a different parser (front-end).
On other machines (e.g. 68000), the compilers use the same front
end oas on the 8086, but share a new back-end.

More information when the new compiler arrives!

Incidentally, before people write to me and say
"why didn't you use another compiler?"
we needed one that could access more than 64K  of data and
would run under Concurrent CP/M.
We couldn't wait for the multi-tasking version of MS-DOS,
which is not available yet, or for the Lattice or CI versions of C
which are not out for CP/M with extended data.
If you know of one that meets these requirements, let me know.

Mike Ciaraldi
ciaraldi@rochester