From clive@on-the-train.demon.co.uk Fri Feb 7 16:56:45 1997 Received: from relay-11.mail.demon.net (relay-11.mail.demon.net [194.217.242.137]) by dkuug.dk (8.6.12/8.6.12) with SMTP id QAA18778 for ; Fri, 7 Feb 1997 16:56:44 +0100 Received: from on-the-train.demon.co.uk ([158.152.83.191]) by relay-9.mail.demon.net id aa911636; 7 Feb 97 14:45 GMT Message-ID: <2v3vppAIdz+yEwVy@on-the-train.demon.co.uk> Date: Fri, 7 Feb 1997 14:08:08 +0000 To: sc22wg14@dkuug.dk From: "Clive D.W. Feather" Reply-To: clive@demon.net Subject: Access to objects - N640 MIME-Version: 1.0 X-Mailer: Turnpike Version 3.03 beta 2 <81yImECxEkLwWotvkdN7a29E6a> I endorse this proposal. However, I would like to suggest a change. The fifth bullet point covers access to an object that is part of a larger object. As worded, this implies that I can access any int variable using any structure type, such as: struct s { double d; int i; } s; int a [sizeof s]; s.d = 1.0; s.i = 2; * (struct s *) a = s; [Okay, in this case the cast might be dubious, but I can write something similar using malloc.] I don't think this is what we want. Proposal: replace the fifth bullet point with: if an aggregate or union type, a type whose element or member types (including recursively element and member types of subaggregates and contained unions) are related to the element and member types of the effective type of the object in one of the above ways, or Replace the full stop on the last bullet point with a comma, and add at the end: or as part of an access, conforming to the above requirements, to an enclosing object whose effective type is an aggregate or union type. There's no need to mention recursion, because you simply apply this rule to the outermost object. And there's no need to have the "that includes" bit, because the access will be forbidden by "the above requirements". The present footnote 43 should be expanded to something like: This does not mean that an object includes type information within itself. Rather, it allows an implementation, by examining the effective types of expressions, to determine when an object can or cannot be aliased. -- Clive D.W. Feather | Associate Director | Director Tel: +44 181 371 1138 | Demon Internet Ltd. | CityScape Internet Services Ltd. Fax: +44 181 371 1150 | | Written on my laptop - please reply to the Reply-To address