[Cfp-interest] updates for Parts 3

[Jim Thomas]

On Mar 20, 2013, at 9:35 PM, Fred J. Tydeman <tydeman at tybor.com> wrote:

> On Thu, 14 Mar 2013 21:22:13 -0700 Jim Thomas wrote:
>> 
>> I posted updated drafts for TS 16881 Part 2 (n1684) and Part 3 (n1685)
> 
> Some comments.
> 
> PDF page 7, lines 28-34, are a bigger font than other lines.

Right.

> 
> PDF page 8, line 26:  'incudes' -> 'includes'

Right.

> 
> PDF page 13, line 8 has: "Extended floating types are not compatible
> with any other types."  So, what if 'long double' is the Intel x87
> 80-bit format?  Would it not be compatible with _Float64x?

They might have the same formats and all the same operations, but would not be "compatible" in the C sense.

> 
> PDF page 16: Do we need to add: FLTn_HAS_SUBNORM, FLTnX_HAS_SUBNORM,
> DECn_HAS_SUBNORM, DECnX_HAS_SUBNORM [before paragraph 13b]?  Can their
> values change depending upon FP control word settings?  My
> understanding is Intel and ARM have a FP control word bit to
> enable/disable flush denorms to zero [for binary FP] (which is not
> part of Standard C, but could be an extension).

I believe these macros should be determinable at compile time. If not, then they should have been functions, say getunderflowmode() (which I'm not proposing).

> 
> PDF page 20, last line: Mentions INFINITY and NAN, yet I do not see
> them.  I only see HUGE_VAL_* and SNAN*.

DEC_INFINITY and DEC_NAN are defined in Part 2. Maybe that should be mentioned.

> 
> PDF page 21, lines 28-29.  Do we need to add [dec/bin]encodingdNx_t
> for the extended types?

The extended types are not intended for data exchange and IEC 60559 doesn't specify their encodings.

> 
> PDF page 28, line 41.  Need blank line between fdim* and fmax*.

Right.

> 
> PDF page 30, lines 17-21.  Why no Nx versions?

They should be there. IEC 60559 requires them for all arithmetic formats.

> 
> PDF page 30, lines 39-40> Why those functions for data-interchange
> types?  Is it a requirement of IEEE-754?

It's a recommendation (should) of IEC 60559.

-Jim

> 
> 
> ---
> Fred J. Tydeman        Tydeman Consulting
> tydeman at tybor.com      Testing, numerics, programming
> +1 (775) 287-5904      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