J3/06-274r1 Date: 14 August 2006 To: J3 From: Dick Hendrickson Subject: typos and editorial suggestions for INTRO to clause 4, 6, and 12 References: J3/06-007 Comments that I believe are merely editorial and have no technical impact or merit. Editorial comments on MACROS are in the macro paper, not this one. General edit questions 1) Is there a way to add the defining rule number in the BNF for things that are far away. For example R448 component-initialization is initialization-expr and initialization-expr is defined way off in C 7. Withdrawn, no action proposed 2) In some/many cases Rules do not have a blank line between them, whereas constraints always seem to. Makes it hard to read the rules when they are close together. See, e.g., [120:13-22] Straw vote: all the same? ad hoc? Specific edits XV The laundry list needs to be redone at a later time. I suggest going down the list of things in spread sheet and having a big feature and a little feature list. Add J3 note for future action [1:12] Add after "length" "; part 2 has been largely subsumed by allocatable characters with deferred length parameters." [2:40] change "(16)" to "(2.2)" [3:15] This seems to say the new .XOR. has a different precedence from what it had before. It didn't have one before. Withdrawn, deferred to paper 06-212 [6:2] Clause 4 has constraints with 3 digits. Edits: [6:2] Replace "Csnn" with "Csn" [6:3] Replace "nn is a two digit" with "n is a two digit or three digit" Van suggest the following clause 4 related edits: [65:17] Replace "POINTER or" with "Neither POINTER nor" [76:34] Append to the end of the sentence "to that name" [12:7] To my eye on my printer, this line looks like it has a smaller font than the others. Withdrawn, no action [12:6] Change "are to be" to "may be". Also in [13:25]. The current wording sounds like they must be available, but availability is controlled by a USE statement. Withdrawn [13:34] "the ancestor" implies to me the immediate unique ancestor. Can't it be declared in any ancestor? Withdrawn [15:13] change "a procedure" to "a Fortran procedure". [21:21] There are two instances of "may be accessed". The first is really a requirement, the second is mostly expressing an option. Replace the first clause to refer only to co-array elements on other images and say "may only" and use some form of "may or may not" on its own image. Withdrawn [41:22] Delete the phrase "considered to be" which implies that it's a matter of opinion. Editorial suggestion: Check all cases of "considered to be" and "said to be" for deletion or rewording. [42:note 4.1] Delete the phrases "by design" and "are designed to be" because they imply doubt. In line one of the note, change "is known" to "is always known". In the last line, change "Generics ... to be" to "All generic references are" [44:1] Add a forward reference for extensible type, "(4.5.7)" [44:30] Add a note explaining that if A is type compatible with B, that doesn't imply that B is TC with A Insert after [44:30] as a note "Given TYPE TROOT ... TYPE,EXTENDS(TROOT) :: TEXTENDED ... CLASS(TROOT) A CLASS(TEXTENDED) B ... A is type compatible with B by B is not type compatible with A." [53:3] The processor only defines the collating sequence for some character sets. ISO_10646 and ASCII are well defined. The default character set sequence is partially defined. This also (possibly) affects [52:22-23] and the text that should be moved here from [25:16-21] Withdrawn [53:note 4.14] My Japanese is a little rusty, does this translate to "always buy the NAG compiler"? Add at the end of the note "This translates to 'Without her, nothing is possible'". [54:26] Delete this line. It's the only sub header under bits and none of the other types have one like this. [54:28-29] Bold font "kind" and "representation method". All of the other types have theirs bolded. [54:32] For bits, it would be user friendly to add a note saying that they use "kind" more like the character length parameter and people shouldn't get confused. Particularly because the standard is weak at keeping separate the distinction for characters. I'm sure I've seen things like "ASCII type". Withdrawn [61:note 4.27] I'd expand on the explanation by saying that length type parameters don't participate in generic resolution, only the KIND ones do. And that a KIND parameter can (also) be used as in a length-like way, but not vice-versa. Withdrawn [64:note 4.34] Add, before the line declaring SYNOPSIS PROCEDURE(printer_interface),POINTER :: PRINT => NULL() And, after "TITLE," in the text, insert "a procedure pointer, PRINT, with an explicit interface the same as printer_interface, " [57:6] Add a forward reference for extensible type withdrawn [65:27] Need "and" before "does not". Note to editor: Should R449 and C462 from page 65 be moved to page 84:23+ where initialization is discussed (R505) and text 65:5-7 and 65:26-27 be moved to 85:15+ ? [115:9] Delete "The" [115:12] Delete "A" [126:16] Add " but its parameters must be defined" at the end of the sentence. [131:note 6.31] In the penultimate line add "if it was allocated" before the full stop. [330:7-12] Could we change "separate" to "sep" or vice versa in the linked bnf terms? Defer to /B group