COBOL compilers

Trevor,


I read your request for information on COBOL compilers in comp.computers
with some interest. Many years ago I headed up a COBOL compiler project
for NCR for the then-new VRX systems, which included a "COBOL Virtual
Machine" object code set. We produced a "tagged architecture" machine
COBOL as a low-level machine language. (My thesis research on a SNOBOL
machine contributed some features to the product.) I wrote this up in a
paper for the 1978 Fall Joint Computer Conference ("The Criterion COBOL
Compiler"). It discusses the approach we took and some of the problems we
ran into.


The compiler itself was relatively straight-forward, much like a text-book
example. (The fact that I was teaching compiler construction at San Diego
State University at the time may have influenced my approach.) We
compiled the DATA DIVISION statements into descriptors which were handled
by the runtime firmware.


If you find the article, read it, and have any interest in discussing any
of our approaches, don't hesitate to ask.


Incidentally, we were one of the first COBOL'74 compilers to go through
the Federal COBOL Compiler Validation Service testing. They found only
four problems: two console I/O interpretation questions and two incorrect
level diagnostics. My COBOL background before coming to the project was
one program in a programming languages course. I later wrote a couple of
more programs as tests, which made me an expert among most computer
scientists! I still have a reading knowledge of the language.


                                                  Michael
--
Michael D. Shapiro, Ph.D. e-mail: mshapiro@nosc.mil
NRaD (NCCOSC RDT&E Division) Code 411 San Diego CA 92152-5000
Voice: (619) 553-4080 FAX: (619) 553-4808 DSN: 553-4080
  [Until January 1992 we were Naval Ocean Systems Center (NOSC)]
--

Return to the comp.compilers page.
Search the comp.compilers archives again.