Re: History and evolution of compilers

Peter Ludemann <ludemann@inxight.com>
8 Oct 1997 00:43:21 -0400

          From comp.compilers

Related articles
Re: History and evolution of compilers mathesonh2@verizon.net (Hal Matheson) (2010-04-02)
History and evolution of compilers pt93mhe@student.hk-r.se (Martin Hellspong) (1997-09-30)
Re: History and evolution of compilers wclodius@lanl.gov (William Clodius) (1997-10-01)
Re: History and evolution of compilers ela@fluxion.hut.fi (Eero Lassila) (1997-10-01)
Re: History and evolution of compilers bill@amber.ssd.csd.harris.com (1997-10-01)
Re: History and evolution of compilers mkent@acm.org (Mike Kent) (1997-10-02)
Re: History and evolution of compilers henry@zoo.toronto.edu (Henry Spencer) (1997-10-02)
Re: History and evolution of compilers ludemann@inxight.com (Peter Ludemann) (1997-10-08)
Re: History and evolution of compilers cbbrowne@hex.net (1997-10-10)
Re: History and evolution of compilers mark@hubcap.clemson.edu (1997-10-10)
Re: History and evolution of compilers norman@kbss.bt.co.uk (Norman Hilton) (1997-10-10)
Re: History and evolution of compilers rweaver@ix.netcom.com (1997-10-14)
Re: History and evolution of compilers mslamm@pluto.mscc.huji.ac.il (1997-10-14)
Re: History and evolution of compilers preston@tera.com (1997-10-16)
[3 later articles]
| List of all articles for this month |

From: Peter Ludemann <ludemann@inxight.com>
Newsgroups: comp.compilers
Date: 8 Oct 1997 00:43:21 -0400
Organization: InXight Software, Inc. http://www.inxight.com
References: 97-09-130 97-10-008 97-10-017
Keywords: Fortran, history

The moderator wrote:


> Fortran IV did get two compilers on the 360 series, Fortran G which was
> fast and generated rotten code, and Fortran H which produced very good
> code. -John]


As I recall, the Fortran G code wasn't "rotten"; in most cases it was
acceptable (maybe you're thinking about the WATFOR/WATFIV compilers).
Wirth's Algol-W compiler produced code of about Fortran G's quality;
and it ran as fast as the WATFIV compiler.


As to Fortran H, the code was good, but we were warned to check our
results because some of the optimizations could break working code
(how much of this was due to people not adhering to the letter of the
Fortran language spec and how much was due to Fortran-H taking
liberties, I don't know). By the way, the "G" and "H" refer to the
size of machine needed for the compilers; if memory serves me, "G" was
around 128K and "H" was 256K or 512K.


--
Peter Ludemann +1.650.813.6806 (fax: +1.650.813.7400)
Software Architect ludemann@inxight.com
Inxight Software, Inc. http://www.inxight.com
PAHV 118, 3400 Hillview Ave., Palo Alto, CA 94304
[It was acceptable, but it was definitely rotten. For example, it made
no attempt to do base-offset branches, but rather loaded an address word
from memory before every branch. Incidentally, IBM apparently bought
Fortran G from someone else, since DEC's PDP-6/10 Fortran compiler used
the same front end with data structures called "rolls". -John]




--


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.