Re: gnubol: Compiler Limits

[RKRayhawk@aol.com]

In a message dated 12/27/99 2:31:14 PM EST, tej@melbpc.org.au writes:

<< - 2**31-1 (otherwise). (even maybe variable names, with
 option/default for warning if you go over 30/31)
  >>
That brings up an interesting issue. At the base we are dealing with C. A 
rough approximation of limits of 2**31-1 instantly pushes count fields and 
comparands to four byte integers (unless of course, being COBOL afficionados, 
we will store these limits in decimal fields :-) ). This brush stroke has 
implications. 

There is a small issue here as well about hardware assumptions. We may wish 
to impose the requirement that our backend compiler can deal in 32 bit ints.  
But that does not mean
that we can assume that the hardware is actually 32 bits. Surely such is 
becoming so common as to be readily characterized as dominant. But do we want 
this compiler's counters
and comparands to be 32 bit carte blanche, no questions asked?

A generalized approach to limits also means that we abandon any hope of 
dealing with parts of the requirements with fixed structures. Every thing 
becomes linked lists.  That is not necessarily good.

We may want to consider the limits one by one. Also we are after the codebase 
which is practical, not a theoretical set of programs. The case structure 
(EVAL), search, sort, and multiple file keys are points of real potential 
performance problems, just in the way we try to compile the material. COBOL 
relates mostly to resource management, not limit exploration. We may be 
justified in establishing limts in certain critical areas.


Best Wishes
Bob Rayhawk
RKRayhawk@aol.com


--
This message was sent through the gnu-cobol mailing list.  To remove yourself
from this mailing list, send a message to majordomo@lusars.net with the
words "unsubscribe gnu-cobol" in the message body.  For more information on
the GNU COBOL project, send mail to gnu-cobol-owner@lusars.net.