To: J3 J3/02-115r1 Subject: Comments on Section 12 From: Subgroup C Date: 27 February 2002 Edits refer to 02-007. Section 1 Edits ---------- 249:7 [Misplaced constraint. Don't worry about these if 02-107 passes, because it deletes abstract interfaces. Otherwise, editor: Delete C1211.] And add a new constraint 244:25+ C1203.5(R1203) If the \si{interface-stmt} is INTERFACE PROCEDURE(), the \si{function-name} in the \si{function-stmt} or the \si{subroutine-name} in the \si{subroutine-stmt} shall not be the same as a keyword that specifies an intrinsic type. ----------- 245:1 [If procedure pointers aren't a problem, it's difficult to imagine how internal procedures could be. However, they are a problem if they are used in DTIO because an internal procedure could be invoked out of context. Editor: change "." to ", or an internal procedure if the generic-spec is not a dtio-generic-spec" ] ------------ This item was 245:5 [PROCEDURE statements don't have generic identifiers. Note: this item was handled in 02-144 Item 6 ] ------------ 251:1- [In lines 3-4 of Note 12.16, the term ``would have be'' is used. This isn't used elsewhere (that I could find), and isn't defined. Editor: ``would have'' becomes ``must'' (yes ``must'' not ``shall'' because it's an ``inevitable consequence of provisions of this standard'' -- or at least that's what the current note implies).] ------------ 249:14+ [Editor: Delete ``section'' in Note 12.12.] ------------ 267:2-3 [Editor: Set ``, or an asterisk'' in obsolescent font.] ------------ 273:1 [I changed ``\si{target}'' to ``\si{data-target} or \si{proc-target}'' at meeting 159. But the ``or \si{proc-target}'' part is irrelevant -- we aren't later going to assign a value to a procedure by way of a procedure pointer. Editor: Delete it.] ------------ 273:1+ [Missed a fine point of pointer components. Does this need an interp?] Add a new item after item 2: item 2.5: As the \si{expr} corresponding to a component with the POINTER atttribute in a \si{structure-constructor}. ------------ 274:3 [Editor: Put a \tdef around ``elemental procedure''. This will embolden it, and put it in the index.] Section 2 Interop question The answer to the question below is "Yes, it is OK." No edits needed. 264:15-17,267:1-3 Is it OK for a DTIO procedure to have the NAME= specifier? If it's not OK, we need to say so about the procedure, not the interface body -- because an interface block with a \st{dtio-generic-spec} might contain a PROCEDURE statement that refers to a module procedure, a dummy argument, a procedure pointer, an external procedure having an interface specified by another PROCEDURE statement, or (if the edit for 245:1 above is accepted) an internal procedure.