To: J3 Members J3/16-266r2 From: Van Snyder & Steve Lionel Subject: Comments on Clause 13 References: 16-007r2 Date: 2016 October 12 1. Edits -------- [277:7-8 13.7.2.3.8p2 Input/output rounding mode] Replace "decimal value" with "decimal or hexadecimal value" twice. [280:16 13.7.6p3 User-defined derived type editing] Replace "that derived type" with "the derived type of the item". [281:6-7 13.8.1.2p1 T, TL and TR editing] Replace "the left tab ... of that record" with "character position one of that record becomes the left tab limit" {because this is easier to parse (and shorter) than the existing inside-out phrase}. [285:24+1-6 NOTE 13.31 Null values] Replace "All blanks in a list-directed input record" with "All blanks encountered during list-directed input". (The term "list-directed input record" is poorly specified, and leaves ambiguous what happens when list-directed input starts in the middle of a record as positioned by a prior nonadvancing input.) [289:29 13.11.3.3 Namelist input values] Before the sentence beginning "Such a sequence", insert the sentence "The delimiters may be apostrophes or quotes; the value of the , if any, in the for an is ignored." [289:34+ NOTE 13.36 13.11.3.3 Namelist input values] Delete the first and third sentences of this note, which read as normative text; they have been inserted in the previous edit. The note remains with the sentence "The delimiter is required..." [290:2 13.11.3.4p1(3) Null values] Change the third bullet from: "zero or more blanks preceding the first value separator and following an equals, or" to: "a value separator that is the first nonblank character following an equals, or" [291:21 13.11.4.3p3 Namelist output records] Replace "will be produced by namelist formatting" with "is placed". [291:23 13.11.4.3p3 Namelist output records] Replace "is produced by namelist formatting" with "is placed in the output record". 2. Optional edits at the editor's discretion -------------------------------------------- [273:27 13.7.2.3.3p1 E and D editing] [274:12 13.7.2.3.4p2 EN editing] [275:8 13.7.2.3.5p2 ES editing] Replace the comma before "otherwise" with a semicolon. 3. Rejected edits ----------------- [267:20-21 C1306] Replace the constraint: "C1306 (R1308) shall be positive for L or A edit descriptors [and may be positive or zero for all other edit descriptors]. {Because the syntax for doesn't allow a sign. (This adds no value.) [269:22-32 13.4p9] Replace "in a manner identical to the way it is positioned when" with "as if by"; delete "is processed". {Compare to [278:35 13.7.4p5].} (The text on 278 adds a much longer description of the process that would be necessary to repeat here. The current wording is fine.) [270:7-8 13.5p4] Replace "in a manner identical to the way it is positioned when" with "as if by"; delete "is processed". {Compare to [278:35 13.7.4p5].} (See above.) [271:15+ 13.7.2.1p1(3)] Insert a note: "NOTE 13.7a The effect of the scale factor is described in 13.8.5." (Nowhere here is the scale factor even mentioned.) [273:3 13.7.2.3.2p7 F editing] Append a sentence: "The scale factor has no effect on hexadecimal input editing." {If it's true. It should also be stated in 13.8.5. Compare to [274:9 13.7.2.3.4p1].} (It is true because hexadecimal input always has an exponent, so is excluded from scale factor, but see 16-284 for a related issue) [276:2- NOTE 13.13+ ES edit descriptor] Insert a note: "NOTE 13.13a The effect of the ESw.dEe edit descriptor is the same as the effect of the 1PEw.dEe descriptors." (While true, such a note doesn't add value.) [276:10 13.7.2.3.6p2] Insert "decimal" before "digits". (Unnecessary - this text is discussing the number of digits - what those digits are is described in line 21.) [277:25 13.7.2.4p1] Append a sentence "The scale factor has no effect on B, O, or Z editing." {If it's true.} (Unnecessary as 13.8.5 doesn't include B,O,Z in the list of edit descriptors it affects) [282:14 13.8.5p1] Insert "only" before "the editing done...". (Unnecessary.) [285:23 13.10.3.2p3] Replace " in the input list becomes" with "s in the input list become" {because there might be more than one of them}. (The existing text already allows for that.) [289:34+1-5 NOTE 13.36] The first and third sentences of NOTE 13.36 appear to be normative, and the first sentence is (improperly) written as normative text. Make the first and third sentences normative, in a paragraph at [289:34+ 13.11.3.3p7+. Leave the second sentence in the note. (Replaced with an edit in section 1.) 4. Questions and comments without edits --------------------------------------- [281:4-7 13.8.1.2p1] The relationship of the left tab limit to a data transfer statement executed after a nonadvancing data transfer statement to the same unit should be explicitly specified. It shouldn't be necessary to prove a theorem involving "immediately prior to nonchild data transfer." (JOR finds the current wording sufficient.) [282:13-30 13.8.5] The relationship of P editing to hexadecimal input during F editing should be described here. The relationship of P editing to a data transfer statement executed after a nonadvancing data transfer statement should be obvious here. Something like "The value of $k$ is zero at the beginning of execution of a [nonchild?] data transfer statement." If that's true. (P editing has no effect on input of a hexadecimal-significand field as such values always have exponents and are thus exempt from scale factors. Paper 16-284 adds words related to this. Your second question is addressed adequately by 12.5.2 paragraphs 3 and 5.) [283:24-29 13.10.2p2] Can r be zero? ("r is an unsigned, nonzero, integer literal constant.") [287:8-13 13.10.4p9-10] If character sequences are both preceded and followed by value separators, null values would be produced. This contradicts 13.10.4p12. (This text is specifically about delimited character sequences, so the delimiters would not create null values.) [287:18+1-2 NOTE 13.33] Is the length of a list-directed output record bounded by the RECL= specification in the OPEN statement? (Yes (229:34), but it could be shorter.) [287:29-38 13.11.2p2] More needs to be said about constraints and restrictions on , especially concerning pointer and allocatable components, whether a pointer component can appear after a with nonzero rank, whether two s with nonzero rank can appear, .... (You're welcome to submit a paper on this topic.) [288:23-289:39 13.11.3.3] Subclause 13.11.3.3 could be substantially shortened and simplified by reference to list-directed input. (Almost but not quite. There are subtle differences, especially for character input. It's not the right time to worry about this.) [289:1 13.11.3.3p1] Can r be zero? (No - 288:35) [290:2 13.11.3.4p1(3)] This appears to imply that input to an array A using "A= ,10" specifies a value for A(3), i.e., there are two null values. The discussion of blanks as null values appears to contradict subclause 13.11.3.5. (This is handled by an edit in section 1.) [291:12+1-2 NOTE 13.40] Is the length of a namelist output record bounded by the RECL= specification in the OPEN statement? (Yes (229:34), but it could be shorter.)