[1000] in testers
Re: discuss built with -g
daemon@ATHENA.MIT.EDU (daemon@ATHENA.MIT.EDU)
Fri Aug 3 15:58:02 1990
Date: Fri, 3 Aug 90 15:57:40 -0400 (EDT)
From: John F Carr <jfc@ATHENA.MIT.EDU>
To: Ken Raeburn <Raeburn@MIT.Edu>
Cc: testers@ATHENA.MIT.EDU
In-Reply-To: <9008031426.AA06388@multics.MIT.EDU>
hc -g -O is only useful if you expect a core dump (in which case the
program doesn't belong in the release). -g disables most optimizations
and causes the compiler to insert after each statement an instruction
which causes the processor to wait for completion of all memory
operations before continuing. hc -g -O generates code similar in
quality to hc with no options.
I know of 3 bugs in hc2.1y code generation. One is an optimizer bug,
the others I am unsure of. I know of no bad interactions between -g and
-O (except that -O is mostly ineffective in this case).
