[Cfp-interest 1612] Re: TS3 annex review

[Jim Thomas]

Fred, thanks for the careful review and comments.

> On Jun 1, 2020, at 10:49 PM, Fred J. Tydeman <tydeman at tybor.com> wrote:
> 
> On Mon, 25 May 2020 17:15:24 -0700 Jim Thomas wrote:
>> 
>> Look for all sorts of errors, technical and editorial. Send comments to me and cc the review team. I'm proposing you complete your review and send comments by Monday, June 8.
>> 
> 
> Comments on Annex X, 2020/05/25 PDF version
> 
> The copyright date in the footer is out of date.
Right.
> 
> I believe "proper subset" is the more common form of "strict subset”.
The C standard uses “strict subset” a few times but never “proper subset”.
> 
> Page 3, both tables:  emax formula: right column: has 2x instead of 2**x
> That is, the 'x' should be a superscript.
PDF rendering problem. See Damian’s email. I believe all the problems of exponent vs subscript and floor and and ceiling looking like absolute value are PDF rendering problems. Will investigate. Please check the DOCX version, which I believe has these correct.
> 
> Page 3, both tables: emin formula:  3-emax would be better.  Some
> people interpret -emax+3 as -(emax+3).  The same change would apply to
> other uses of -emax+3 in the document.
I’m ok with the change.
> 
> Page 6, line 36: Should DEC_EVAL_METHOD 0 have words similar to 1 (eg,
> subset of values of _Decimal32).
Not needed. There are no decimal types narrower than _Decimal32.
> 
> Page 8, line 3, 20:  Is that absolute value symbol or ceiling symbol?
PDF rendering problem.
> 
> Page 8, line 8:  Is that absolute value symbol or floor symbol?
PDF rendering problem.
> 
> Page 8, line 20:  Is emin-1 supposed to be the exponent of 2?  
PDF rendering problem.
> Since
> log(a**b) is b*log(a), it might be clearer to use: ceiling((emin-1)*log10(2)))
We’re using the same form as the C standard uses for FLT_MIN_10_EXP.
> 
> Page 8, line 32, 36:  Is emax supposed to be an exponent?  Is -p supposed
> to be an exponent?
PDF rendering problem,.
> 
> Page 8, line 43:  Is 1-p supposed to be an exponent?
PDF rendering problem.
> Also, add "normalized" between "least" and "value”.
Not needed because all these types have IEC 60559 formats. I think adding “normalized” would be more confusing than helpful.
> 
> Page 8: Need to add after FLT*MAX, a new section: FLT*_NORM_MAX (to be
> consistent with C2X).
Hmm. The NORM_MAX macros aren’t needed for types with IEC 60559 formats. And (draft) C2X doesn’t have them for decimal types (which all have IEC 60559 formats).
> 
> Page 9: line 1: Is emin-1 supposed to be an exponent?
PDF rendering problem.
> 
> Page 9: line 7: Is emin-p supposed to be an exponent?
PDF rendering problem.
> 
> Page 12, line 14:  Would it be better to replace (in C2X) "_Decimal32,
> _Decimal64,  _Decimal128" with "_DecimalN, where N is multiple of 32”.
Why? Only 32, 64, and 128 pertain.
> 
> Page 18, line 3:  Should "M<=N" be "M<N”?
No. There are narrowing functions with _FloatNx parameters that return _FloatN results.
> 
> Page 29, line 42: Should *exp be a superscript?
PDF rendering problem.
> 
> Page 29, line 44: Should exp be a superscript?
PDF rendering problem.
> 
> Several places:  "applys" should be "applies”.
Right.
> 
> Page 32, line 32:  "delare" should be "declare”.
Right.

- Jim Thomas

> 
> 
> 
> ---
> Fred J. Tydeman        Tydeman Consulting
> tydeman at tybor.com      Testing, numerics, programming
> +1 (702) 608-6093      Vice-chair of PL22.11 (ANSI "C")
> Sample C99+FPCE tests: http://www.tybor.com
> Savers sleep well, investors eat well, spenders work forever.