Path: utzoo!mnetor!uunet!husc6!mailrus!ames!lll-tis!daitc!jkrueger
From: jkrueger@daitc.ARPA (Jonathan Krueger)
Newsgroups: comp.databases
Subject: Re: Quel vs SQL in Ingres 5.0
Message-ID: <44@daitc.ARPA>
Date: 12 Apr 88 00:15:11 GMT
References: <4350@ihlpf.ATT.COM> <551@hscfvax.harvard.edu>
Reply-To: jkrueger@daitc.UUCP (Jonathan Krueger)
Organization: Defense Applied Information Technology Center, Alexandria VA
Lines: 31
Keywords: quel, sql, ingres

G.Pavlov writes:
>...My own view is that the [SQL] portability issue is overexaggerated...

Two quotes from C. J. Date's excellent "A Guide to The SQL Standard"
(Addison-Wesley, 1987) seem relevant:

p. 3  "In many ways, the SQL standard is not particularly useful in
itself; it has been characterized, perhaps a little unkindly, as `the
intersection of existing impelmentations,' and as such is severely
deficient in a number of respects...there are some fifty or so SQL
implementations available today, no two of those implementations are
precisely identical, and none of them is precisely identical to
standard SQL!  Even the IBM implementations in SQL/DS and DB2 are not
100 percent compatible with each other, and each of them differs from
System R SQL and also from standard SQL on numerous points of
detail--not all of them trivial, incidentally."

p. 5  "SQL in particular is very far from ideal as a relational
language...in some important respects, it fails to realize the full
potential of the relational model..although there are well-established
principles for the design of formal languages, there is little evidence
that SQL was ever designed in accordance with any such principles.  As
a result, the language is filled with numerous restrictions, ad hoc
constructs, and annoying special rules.  These factors in turn make the
language hard to define, describe, teach, learn, remember, apply, and
implement....Standard SQL leaves as `implementation-defined' certain
aspects that would be much better spelled out as part of the
standard...as a result, it seems likely that every realistic
implementation of the standard will necessarily include many
implementation-defined extensions and variations, and hence that no two
`standard' SQL implementations will ever be truly identical"