home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.edu:1666 comp.lang.fortran:3562 comp.lang.misc:3083 comp.arch:9426 sci.math:11484
- Newsgroups: comp.edu,comp.lang.fortran,comp.lang.misc,comp.arch,sci.math
- Path: sparky!uunet!utcsri!helios.physics.utoronto.ca!alchemy.chem.utoronto.ca!mroussel
- From: mroussel@alchemy.chem.utoronto.ca (Marc Roussel)
- Subject: Re: User-Defined Operators (was: Scientists as Programmers (was: Small Language Wanted))
- Message-ID: <1992Sep15.154932.5907@alchemy.chem.utoronto.ca>
- Followup-To: comp.lang.misc
- Organization: Department of Chemistry, University of Toronto
- References: <1992Sep14.090213.18722@uklirb.informatik.uni-kl.de> <mcdonald.299@aries.scs.uiuc.edu> <9225915.145@mulga.cs.mu.OZ.AU>
- Distribution: na
- Date: Tue, 15 Sep 1992 15:49:32 GMT
- Lines: 37
-
-
- Note that I have redirected followups to comp.lang.misc. I fail to see
- the relevance of this discussion to the other newsgroups mentioned;
- certainly it has little to do with mathematics.
-
- In article <9225915.145@mulga.cs.mu.OZ.AU> fjh@munta.cs.mu.OZ.AU
- (Fergus James HENDERSON) writes:
- >mcdonald@aries.scs.uiuc.edu (J. D. McDonald) writes:
- >>kirchner@uklira.informatik.uni-kl.de (Reinhard Kirchner) writes:
- >>>Take Pascal-XSC. It is available on all unix machines since it compiles to
- >>>standard C. It allows users to define operators monadic and dyadic, and it
- >>>comes with many of them for interval, vector, matrix, complex etc.
- >>>computations.
- >>
- >>However, the results will be no faster .... and almost surely slower,
- >>than simply coding in C.
- >
- >Why "almost surely slower" than C?
-
- Automatic source translation to a high-level language almost never
- generates code that a human would have written. Since compilers and
- optimizers are tested almost entirely with human-written code, it is
- very likely that machine-translated code will break optimizer
- assumptions. At best, this means that the code will run more slowly
- than a program written directly in the eventual compilation language; at
- worst, this exposes compiler and optimizer bugs which a compiler vendor
- will be reluctant to look at due to the obtuseness of the code.
- For an example of this sort of thing, consult the June 1992 issue
- of the Hewlett-Packard journal. Their Fortran optimizing preprocessor
- (which does Fortran->Fortran source translation) uncovered some bugs in
- both the compiler front end and in the optimizer for the reasons stated
- above.
-
- Sincerely,
-
- Marc R. Roussel
- mroussel@alchemy.chem.utoronto.ca
-