compiling GNU Emacs 18.51's data.c -OG -v on Pyramid OSx 4.0

Bob Sutterfield bob at allosaur.cis.ohio-state.edu
Fri Jul 22 07:03:06 AEST 1988


Environment:  GNU Emacs 18.51, Pyramid OSx 4.0

This isn't really a bug in Emacs, but it's pretty interesting
nonetheless, probably to those same people who worry about Emacs and
compiler development...

I'm finally getting around to installing GNU Emacs 18.51 on our
Pyramids under a heavily-PTF'd OSx 4.0.  When I install things on our
Suns, I use `gcc -O -gg', but (alas) I have no gcc for a Pyramid, so I
compile things like `cc -OG -v' which invokes the Pyramid cc's global
optimizer and instructs it to brag verbosely at me about how many
statements it found that could be moved outside loops, and the like.

I find that when I get around to compiling src/data.c, the C compiler
loops indefinitely in /lib/ccom.  data.c compiles appropriately
quickly with simply -O, but cc -OG runs for a very long time.  No, I
haven't waited for more than a few hours to see if it ever terminates.

Has anyone else observed this with any other optimizing compilers?
Pyramid, is this an occasional feature of your global optimizer - that
it gets so zealous about its work that it can't decide when to quit
slicing and dicing the code?  Is there a particular chunk of data.c
that could be slightly modified so that it would pass safely through
optimizers that have such features?
-=-
 Bob Sutterfield, Department of Computer and Information Science
 The Ohio State University; 2036 Neil Ave. Columbus OH USA 43210-1277
 bob at cis.ohio-state.edu or ...!{att,pyramid,killer}!cis.ohio-state.edu!bob



More information about the Comp.sys.pyramid mailing list