From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me
X-Spam-Level: 
X-Spam-Status: No, score=-1.3 required=5.0 tests=BAYES_00,INVALID_MSGID
	autolearn=no autolearn_force=no version=3.4.4
X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: 103376,34c128c452eed637
X-Google-Attributes: gid103376,public
From: kilgallen@eisner.decus.org (Larry Kilgallen)
Subject: Re: How is an ADA compiler done?
Date: 1996/10/30
Message-ID: <1996Oct29.213612.1@eisner>#1/1
X-Deja-AN: 193188618
x-nntp-posting-host: eisner.decus.org
references: <54u38v$g9o@mikasa.iol.it> <dewar.846423013@merv>
 <555e28$v6c@mikasa.iol.it>
x-nntp-posting-user: KILGALLEN
x-trace: 846642976/24752
organization: LJK Software
newsgroups: comp.lang.ada
Date: 1996-10-30T00:00:00+00:00
List-Id: <comp.lang.ada>


In article <555e28$v6c@mikasa.iol.it>, schizophonic@iol.it writes:
> dewar@merv.cs.nyu.edu (Robert Dewar) wrote:
> 
>>Are atrtibute grammars used? Yes by one or two compilers, No by most. I don't
>>think it is a particularly helpful formalism. At least one commercial compiler,
>>the DDCI compiler does, as far as I know use this approach, but on the other
>>hand, it is much later than expected, so it is certainly not a demonstration
>>that this technique is specially effective (it is also not a demonstraion
>>to the contrary either, there are MANY things that can make compilers later
>>than hoped, as all of us in the business know well!)
> 
> Sounds strange; in my teacher's opinion, nowadays about the 50% of
> commercial compilers are based on attribute grammars; in facts the
> formalism was developed from the state of the art in the compiler's
> algorithm (at least, the state of the art when Knuth started his
> work). Is Ada (now I wrote it exactly) an exception, or is my teacher
> wrong?

Whether Ada is an exception by having only "one or two" compilers
developed that way depends on how lenient your teacher is with the
phrase "about 50%".  It is quite possible, however, that there are
other languages biased in the other direction, so "about 50%" may
be true across all languages.

The quality of a given compiler is not necessarily related to
whether "state of the art" techniques for arbitrary compiler
construction are utilized.  In fact, even accurate rendition of
the language is not the sole criterion for a quality compiler.
I believe it was Knuth, in fact, who developed the Web language
on the basis that _documentation_ of the implementation was one
of the most important aspects of a program.  For some situations
quality of the error messages may be most important, and I doubt
whether that goal is best served by formal approaches.

Larry Kilgallen