From owner-sc22wg5 Wed Jul 17 19:04:36 2002 Received: from nameserv.rl.ac.uk (nameserv.rl.ac.uk [130.246.135.129]) by dkuug.dk (8.9.2/8.9.2) with ESMTP id TAA54253 for ; Wed, 17 Jul 2002 19:04:36 +0200 (CEST) (envelope-from jkr@jkr.cc.rl.ac.uk) Received: from jkr.cc.rl.ac.uk (jkr.cc.rl.ac.uk [130.246.8.20]) by nameserv.rl.ac.uk (8.8.8/8.8.8) with ESMTP id SAA01852 for ; Wed, 17 Jul 2002 18:05:05 +0100 Received: (from jkr@localhost) by jkr.cc.rl.ac.uk (8.8.8+Sun/8.8.8) id SAA01815 for SC22WG5@dkuug.dk; Wed, 17 Jul 2002 18:08:40 +0100 (BST) Date: Wed, 17 Jul 2002 18:08:40 +0100 (BST) From: John Reid Message-Id: <200207171708.SAA01815@jkr.cc.rl.ac.uk> To: SC22WG5@dkuug.dk Subject: Re: Edits to the draft standard MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Here is a revised version of the paper I sent yesterday. Please discard yesterday's version. Some of my edits represent small technical changes to the standard. Next time round, I will separate them off into another paper. Comments on what is here, please. John. ............................. 02-xxy To: J3 From: John Reid Subject: Major edits Date: 17 July 2002 I would like to acknowledge the help of Malcolm Cohen, Richard Maine, and Rob James with the construction of some of these edits. 42:7. Change 'extensible type' to 'extensible type that has no type parameters'. [This brings the constraint into line with the text in 4.6, which says that an alias is not a new type. If we were to allow extension of an alias to a type with parameters, we would be treating it as a new type that has lost its type parameters and there is no text that explains this.] 43:10. After 'shall' add 'either be an initialization expression or be a specification expression that does'. [The wording in Corrigendum 1 to F95 allows for an initialization expression here. I believe that the present wording will lead to an inconsistency with f95. See also C550 on p. 79.] 43:14. Change 'colon or' to 'colon, be an initialization expression, or'. [See previous edit.] 43:31 to 44:3. Replace C438 and C439 by C438 (R436) If PASS appears, the procedure component shall have an explicit interface specified by . A dummy argument of the interface shall meet the requirements for a passed-object dummy argument (4.5.1.6). It shall be named if appears; otherwise, it shall be the first dummy argument. It shall be polymorphic if and only if type type-name is extensible. All of its nonkind type parameters shall be assumed. [I think it is clearer not to repeat things that are the same. I think it needs to be said that the interface is specified by . Finally, I found it hard to interpret C441 until I realized it was all part of this, so I have included it here,] 44:5. Delete. [Included in new C438.] 44:25&27. Replace '' by '. [There are usually many specific procedures associated with a single generic. The present syntax is not wrong because you can write many GENERIC statements, one for each specific. However, this is unnecessarily clumsy. I at first thought that this constraint was in error because it specifies only one specific.] 45:11-19. Replace C455, C456 by C455 (R440) If PASS appears, a dummy argument of each interface specified by or shall meet the requirements for a passed-object dummy argument (4.5.1.6). It shall be named if appears; otherwise, it shall be the first dummy argument. It shall be polymorphic if and only if type type-name is extensible. All of its nonkind type parameters shall be assumed. [I think it is clearer not to repeat things that are the same. The bit about the interface being explicit is a distraction since it is always the case for bindings and is covered by other constraints. Finally, I found it hard to interpret C462 until I realized it was all part of this, so I have included it here,] 45:31. Delete. [Included in new C455.] 47:19-22. Delete paragraph. [It is wrong, since the first and second sentences are inconsistent for sub-sub-components. It is also incomplete. Why try this, given that the concept has already been defined in the previous paragraph?] 50:4-5. Replace by The interface of a specific binding specified with is that of if present or that of the parent type otherwise. The interface of a specific binding specified with is that of the procedure. The interface of a generic binding contains the interfaces of all the procedures in its s and s and all those of the parent type that have the same and are not overridden. 51:4. After 'specific procedure' add 'that overrides (4.5.3.2) the inherited binding'. 51:5. Delete '(4.5.3.2)'. [We need to make it clear that the rules of overriding apply when the inherited binding is deferred.] 51:14-17. Delete sentence. [I do not think implicit PASS is consistent with the bnf. See for example C458/9 on p. 45. I think both the standard and the user's program will be clearer if PASS is specified whenever it is wanted.] 57:1-3. Change to In a generic binding, each specific interface that has the same name in the parent binding shall satisfy the rules for overriding a specific binding. If there are others, they extend the generic interface and shall satisfy the requirements specified in 16.2.3. 60:4. Change 'target' to ' or ' and delete the internal note. [Reasons are in the internal note.] 81:28. Start a new paragraph after 'pointer.' and change the sentence that follows to 'If a name that has the EXTERNAL attribute is the name of a dummy argument, it is a dummy procedure.' [This change is also needed for the sake of the sentence that follows.] 81:39. Delete. [This is very ambiguous. What does 'above' mean (how far up should I go)? What does 'it' refer to?] 164:11+. Replace the note by C810a (R822) The form shall be used if is not a named variable. [The text on page 165 does not work if there is no and is not named. Also, I took a long time (unsuccessfully) trying to understand exactly what the note meant.] 203:10-11. I do not understand '(or some portion thereof)'. The circumstances when only a portion of the value is used need to be explained. 266:15-18. Delete sentence. [Looks like an early draft of the previous sentence.] 411:34. Change 'name' to 'designator'. [We need to cover variables that are subobjects.]