From willemw@ace.nl Thu Jun 21 18:12:08 2001 Received: from ace.ace.nl (IDENT:root@ace.ace.nl [193.78.104.92]) by dkuug.dk (8.9.2/8.9.2) with ESMTP id SAA05994 for ; Thu, 21 Jun 2001 18:12:06 +0200 (CEST) (envelope-from willemw@ace.nl) Received: from president.ace.nl (IDENT:root@president.ace.nl [193.78.104.250]) by ace.ace.nl (8.9.3/8.9.3) with ESMTP id SAA12884; Thu, 21 Jun 2001 18:12:04 +0200 Received: from president.ace.nl (IDENT:willemw@localhost [127.0.0.1]) by president.ace.nl (8.9.3/8.9.3) with ESMTP id SAA09097; Thu, 21 Jun 2001 18:12:04 +0200 From: Willem Wakker To: "Fred J. Tydeman" cc: "SC22 WG11" Subject: WG11 N478 - Comments from Tydeman on LID In-reply-to: tydeman's message of Wed, 20 Jun 1901 20:17:42 -0500. <200106210122.DAA03374@dkuug.dk> Date: Thu, 21 Jun 2001 18:12:04 +0200 Message-ID: <9094.993139924@president.ace.nl> Sender: willemw@ace.nl Hello Fred, Thanks for your comments. As the comments will be addressed during the revision of LID, I have given the document a WG11 number: WG11 N478. For those who (try to) keep track of WG11 numbered documents: here is the last set of numbers. N471 (SC22/N3203) - Meeting announcement WG11 meeting April 2001 N472 (SC22/N3213) - Liaison Statement from ITU-T SG 7 Regarding ISO/IEC 11404 N473 (SC22/N3244) - Minutes WG11 Meeting April 2001 N474 - LID Roadmap N475 (SC22/N3240) - Response to ITU-T Liaison on 11404 (N472) N476 - LIA-3 for registration ballot N477 (SC22/N3237) - Tables of reply on FDIS 10967-2 N478 - Comments from Fred Tydeman on LID Note that N474 and N476 are to be distributed shortly. Best regards, - Willem Wakker. > Comments on ISO/IEC 11404:1996(E) by Fred Tydeman > > Is it possible to add line numbers (say at every 5th line)? They make > it easier to discuss specific sections. > > Page iv: Missing Annex D Syntax on page 67 (there are two sets of pages > both labeled Annex D and starting on page 67). > > Page 7: 6.3.1 Equality: What about the concept of Not-a-Number (NaN) that > compares unordered with everything, including itself? For those value > spaces that have NaNs, 'a=a' is not true for NaN, so the claim that > 'every' value space has 'a=a' is false. > > Page 8: 6.3.5 Exact and approximate: Is -0 distinguishable from +0? > > Page 9: 6.4 Primitive and non-primitive datatypes: Last sentance: What > is 'linear operators'? Need better examples. > > Page 14: 7.3.3 Character-literal and string-literal: Character-literal > is defined using "'", while string-literal is defined using quote. It > seems that quote should be replaced with '"' (assuming that is correct). > > Page 16: 7.5.2 Dependent values: In Notes 3: "In the same wise" needs > to be corrected to perhaps "In the same manner". > > Page 18: 8.1.1 Boolean: In Note: Change "defined in terms of it" to > "defined in terms of it and NOT". > > Page 21: 8.1.4 Character: In example in note 6: Change ";." to ";". > This may not be the only place with that unusual punctuation. > > Page 24: 8.1.9 Scaled: Redo the line break that splits > in-terger-literal. This may not be the only bad hyphanation. > > Page 24: 8.1.9 Scaled: Note 3: Why is it effectively Rational until > the final result? > > Page 25: 8.1.10 Real: In Values: Change ";." to ";". > > Page 25: 8.1.10 Real: In Notes 2: IEC 559:1988 needs to be updated to > IEC 60559. And, the year needs to be checked. This needs to be done > globally. > > Page 26: 8.1.11 Complex: In Values: Change ";." to ";". > > Page 33: 8.3.1 Choice: Correct spelling of 'operatoins'. In the same > paragraph, what are Or-graph and And-graph? > > Page 43: 8.4.6 Table: In Components: Fix '..' and end of sentance. > > Page 48: 10.1.4 Bit String: In value syntax: Replace 'quote' with '"' > (if that is correct). > > Page 51: 10.1.10 Object identifier: Add "(OSI)" after "Open Systems > Interconnection. > > Page 63: B.3 Procedure-attributes: Redo the hyphanation that split > attributes as at-tributes. > > Page 65: C.3 Floating-Point and C.4 Fixed-Point are inconsistent with > each other with respect to notation. One uses raised periods while the > other uses 'x' as the symbol for multiplication. Raised periods is > better. > > Page 65: C.3 Floating-Point: Correct 'exponent;, and' by removing the > comma. > > Page 67: Annex D: Syntax for the Common Interface Definition Notation: > Fix the 'the the' in the third line. > > Page 70: D.1.10 Real: I believe that the description > 'range(rmin..rmax)' is wrong. It should be along the lines of: > range(-rmax..-rmin) UNION zero UNION range(+rmin..+rmax). Since I did > not see how to do UNION in this notation, there might be missing > functionality in this notation. Both rmin and rmax are positive values. > Need a way to indicate that cannot represent numbers 0 < x < rmin due to > a finite exponent. > > Page 70: D.1.11 Complex: The mapping should be corrected to: > 'only maps value c in C sugh that rmin < |Re(c)| < rmax and > rmin < |Im(c)| < rmax'. > > Page 78: D.5.3 Cyclic enumerated: Correct chaacterizing. > > Page 87: F.8.1.10 Real: Update IEC 559 to IEC 60559. Also, this > paragraph might mention that "scientific numbers" have an unbounded > exponent range, while "floating point numbers" have a finite exponent > range. > > It would help greatly if there were an example mapping showing all three > mappings (not just the "inward" as in Annex E). It would also help if a > more widely known language, such as C, where choosen for the example. > > General comment. I found this a hard standard to follow and I believe > that the various language committees will have a hard time creating a > binding for their language. As I understand, the primary author of > 11404 is also a member of a language committee, yet he has not proposed > a binding for that language (as either an example to put in 11404, or as > a starting point for the language committee to incorporate into their > language standard). > > --- > Fred J. Tydeman +1 (512) 255-8696 Tydeman Consulting > 3711 Del Robles tydeman@tybor.com Programming, testing, numerics > Austin, Texas 78727 Vice-chair of J11 (ANSI "C") > USA Sample C99+FPCE tests: ftp://jump.net/pub/tybor/ > Savers sleep well, investors eat well, spenders work forever. >