[Cfp-interest] nextup()

[Jim Thomas]

> On Aug 19, 2018, at 11:35 AM, Jim Thomas <jaswthomas at sbcglobal.net> wrote:
> 
> 
> 
>> On Aug 2, 2018, at 10:21 AM, Fred J. Tydeman <tydeman at tybor.com> wrote:
>> 
>> Other necks() questions.
>> 
>> Many of the math functions in Annex F in C17, have words like:
>> The returned values are exact and are independent of the current rounding direction mode.
> 
> Agreed.

I suggest the following editorial changes.

Append to F.10.8.5 (TS 18661-1), the sentence:

nextup(x) raises no floating-point exceptions if x is not a signaling NaN. The returned value is independent of the current rounding direction mode.

Append to F.10.8.6 (TS 18661-1), the sentence:

nextdown(x) raises no floating-point exceptions if x is not a signaling NaN. The returned value is independent of the current rounding direction mode.

- Jim Thomas

> Are there other such places?
> 
>> 
>> Perhaps those should be added to necks().
>> 
>> Also, if long double is a pair of doubles, what is nextup(1.L)?
>> Is it
>> 1.L + LDBL_EPSILON
>> or
>> 1.L + LDBL_TRUE_MIN == 1.L + LDBL_EPSILON * LDBL_MIN
> 
> I’d think the latter. The ragged precision of double-double is already a problem in C (e.g., what is p?). and not something CFP needs to address.
> 
> - 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.
>> 
>> _______________________________________________
>> Cfp-interest mailing list
>> Cfp-interest at oakapple.net
>> http://mailman.oakapple.net/mailman/listinfo/cfp-interest
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.oakapple.net/pipermail/cfp-interest/attachments/20180820/ddfc328f/attachment.html