Path: utzoo!mnetor!uunet!lll-winken!lll-tis!ames!sgi!wdl1!mitchell
From: mitchell@wdl1.UUCP (Jo Mitchell)
Newsgroups: comp.databases
Subject: Re: Quel vs SQL in Ingres 5.0
Message-ID: <3560011@wdl1.UUCP>
Date: 15 Apr 88 01:31:11 GMT
References: <4350@ihlpf.ATT.COM>
Lines: 35


	Well, I dunno about all this chatter.  I used Oracle's DB
	for 2 years and Ingres for the last year.  Practically
	speaking I'm generally in favor of SQL.  Why?
	
	(1) No more range variables.
	(2) Not in.

	Range variables, when used properly are ok:  but I've seen
	countless of disjoint queries like:
		range of p is person
		replace p (address = " ") where person.name = "John"
        MAJOR OOPS.  This never happened in SQL. 

	Something else I miss is the "not in" I use to be able to
	use to check whether corrupted data had made its way into
	a table.  IE, a table is only suppose to have values
	in one column that are stored in another table, or only
	addresses for EXISTING employeess should be stored in the address
	table, etc.  Perhaps someone else knows a quick way in QUEL (tell me),
	but the only way I've heard or seen is:
		retrieve into temptable (realtable.all)
		delete temptable where temptable.value = legaltable.values
		retrieve (nr_val = count(temptable.all))
	(in lieu of 
	   select count(*) from realtable where value not in legaltable.values)

	The thing I do like about INGRES is how powerful its application
	building environment is.  True, the forms and report writer aren't
	too hot.  But a similar application that took me 3 mo. in Oracle
	(using FMS) took 1 week in INGRES! True (again), that power in
	the wrong hands can lead to maintenance nightmares.

	Hurray for 6.0!  (When's it on VMS? :>)

							- Jo