Re: Vector assignment semantics (was Re: latest trends in compiler optimization research?)

glen herrmannsfeldt <gah@ugcs.caltech.edu>
Wed, 29 Aug 2007 12:53:56 -0800

          From comp.compilers

Related articles
[11 earlier articles]
Re: Vector assignment semantics (was Re: latest trends in compiler opt gah@ugcs.caltech.edu (glen herrmannsfeldt) (2007-08-17)
Re: Vector assignment semantics (was Re: latest trends in compiler opt Peter_Flass@Yahoo.com (Peter Flass) (2007-08-17)
Re: Vector assignment semantics (was Re: latest trends in compiler opt jjw@cs.sfu.ca (James J. Weinkam) (2007-08-20)
Re: Vector assignment semantics (was Re: latest trends in compiler opt gah@ugcs.caltech.edu (glen herrmannsfeldt) (2007-08-20)
Re: Vector assignment semantics (was Re: latest trends in compiler opt jvorbrueggen@mediasec.de (=?ISO-8859-1?Q?Jan_Vorbr=FCggen?=) (2007-08-21)
Re: Vector assignment semantics (was Re: latest trends in compiler opt tom@kednos.company (Tom Linden) (2007-08-26)
Re: Vector assignment semantics (was Re: latest trends in compiler opt gah@ugcs.caltech.edu (glen herrmannsfeldt) (2007-08-29)
Re: Vector assignment semantics (was Re: latest trends in compiler opt tom@kednos.company (Tom Linden) (2007-09-11)
| List of all articles for this month |

From: glen herrmannsfeldt <gah@ugcs.caltech.edu>
Newsgroups: comp.compilers,comp.lang.pl1
Date: Wed, 29 Aug 2007 12:53:56 -0800
Organization: Compilers Central
References: 07-08-016 07-08-021 07-08-024 07-08-034 07-08-037 07-08-040 07-08-041 07-08-044 07-08-051 07-08-054 07-08-081
Keywords: parallel, PL/I
Posted-Date: 29 Aug 2007 15:56:43 EDT

Tom Linden wrote:


> On Sun, 19 Aug 2007 22:12:19 -0700, James J. Weinkam <jjw@cs.sfu.ca> wrote:


(snip)


>>1. Array expressions are evaluated element by element in row major
>>order, and


>>2. An array assignment is equivalent to a set of nested loops which
>>effectively perform a series of scalar assignments in row major
>>order.


>>This obviates the need for array temporaries except for the final
>>result of an array expression.


(snip)


> We do the same with our (ANSI) compiler.


I thought this started with the note that ANSI didn't do it that way.
That would seem to mean that you have a non-ANSI compiler. (But
probably more compatible with others.)


Note which standards IBM claims for their compiler:


http://publibfp.boulder.ibm.com/cgi-bin/bookmgr/BOOKS/ibmol004/FRONT_1.6


-- glen
[As far as I can tell, the 1976 ANSI PL/I standard says nothing at all about
the order of evaluation of array expressions. But it's very hard to read and
it's possible I missed something. -John]



Post a followup to this message

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