home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!olivea!charnel!sifon!thunder.mcrcim.mcgill.edu!mouse
- From: mouse@thunder.mcrcim.mcgill.edu (der Mouse)
- Newsgroups: comp.lang.c
- Subject: Re: Reasons for using C vs. Fortran or vice/versa
- Message-ID: <1992Nov20.100850.20858@thunder.mcrcim.mcgill.edu>
- Date: 20 Nov 92 10:08:50 GMT
- References: <1992Nov12.135901.15191@ccd.harris.com><1992Nov12.210151.5486@alchemy.chem.utoronto.ca><1992Nov13.171542.10650@taumet.com> <ALANB.92Nov17130727@catalina.sdl.mdcbbs.com>
- Distribution: na
- Organization: McGill Research Centre for Intelligent Machines
- Lines: 28
-
- In article <ALANB.92Nov17130727@catalina.sdl.mdcbbs.com>, alanb@sdl.mdcbbs.com (Alan Braggins) writes:
- [these attribution lines that are confused enough that I'm not sure who
- writes what!]
- >>>>>> On Fri, 13 Nov 1992 17:15:42 GMT, steve@taumet.com (Steve Clamage) said:
- >> mroussel@alchemy.chem.utoronto.ca (Marc Roussel) writes:
-
- >> This is the "aliasing" problem, and dramatically affects the
- >> efficiency of mathematical code in C compared to Fortran.
- > What happened to the proposed "noalias" type qualifier?
-
- It was dropped. It was extremely difficult, perhaps impossible, to
- define its semantics precisely while avoiding, on the one hand,
- weakening it so much as to be useless (to optimizers), and on the
- other, making it so strong as to be useless (to code authors). All the
- proposed definitions of it were sufficiently confusing that even X3J11
- couldn't figure out precisely what they meant. Dennis Ritchie even
- generated a quote which Henry Spencer put into his signature: "noalias
- must go. This is non-negotiable." [from memory - exact wording may be
- wrong].
-
- And sure enough, it was. It's not clear to me what the right solution
- is. I'm sure compiler authors are experimenting with various
- possibilities....
-
- der Mouse
-
- mouse@larry.mcrcim.mcgill.edu
-