[cfp-interest 3623] Re: Rounding Mode

[Jim Thomas]

> On Sep 23, 2025, at 5:21 PM, Damian McGuckin <damianm at esi.com.au> wrote:
> 
> On Tue, 23 Sep 2025, Jim Thomas wrote:
> 
>> Rounding precision mode is in 754-1985, but not in later versions. The attention to rounding precision mode was added to C99 which was designed to support 754-1985. Since 754 no longer mentions rounding precision mode (and assuming the current revision is unlikely to restore it as a feature), why should we elevate its status in C2Y? Just mentioning it as a possible extension, as in footnote 438 in F.8.1, seems appropriate.
> 
> Agreed.
>> 
>> F.8.3 and F.8.4 incorrectly imply that rounding precision mode is an optional feature of ISO/IEC 60559. (Likewise, ?trapping? is no longer mentioned as a user feature in 754. Traps are mentioned only in an implementation note.) We might consider updating F.8.3. and F.8.4 to remove the reference to rounding precision (and trapping) mode which is obsolete from a standards perspective.
> 
> Your call.
> 
>> Choosing terms specifically to accommodate rounding precision mode doesn?t seem needed.
> 
> Sorry. I was not suggesting that. I was unclear.  Silly me.
> 
> The two terms
> 
> 	rounding direction mode
> and
> 	rounding mode
> 
> are used interchangably throughout the standard for exactly the same thing and I am arging that they be made consistent. From what David said, as the IEEE 754 equivalent term is
> 
> 	rounding direction attribute
> 
> it would seem like the former is best?  What do you recommend of the two?

I agree that “rounding direction mode” seems better than “rounding mode”. As previously noted, “attribute” has another meaning in C.

- Jim Thomas

> 
> Thanks - Damian