Re: [PATCH 1/2] system_data_types.7: Add 'fexcept_t'

[Alejandro Colomar <colomar.6.4.3@gmail.com>]

Actually, POSIX already ripped (part of) the description from the C 
standard:

 > Represents the floating-point status flags collectively,
 > including any status the implementation associates with the flags.

This part is in the C standard (and POSIX also has it).

 > A floating-point status flag is a system variable
 > whose value is set (but never cleared)
 > when a floating-point exception is raised,
 > which occurs as a side-effect of
 > exceptional floating-point arithmetic to provide auxiliary information.
 > A floating-point control mode is a system variable whose
 > value may be set by the user to affect
 > the subsequent behavior of floating-point arithmetic.

And this is from POSIX only.

How would you go about it?


 > Represents the floating-point status flags collectively,
 > including any status the implementation associates with the flags.
POSIX describes a
 > [s/A//] floating-point status flag [s/is/as] a system variable
 > whose value is set (but never cleared)
 > when a floating-point exception is raised,
 > which occurs as a side-effect of
 > exceptional floating-point arithmetic to provide auxiliary information.
According to POSIX,
 > [s/A/a/] floating-point control mode is a system variable whose
 > value may be set by the user to affect
 > the subsequent behavior of floating-point arithmetic.

Like this?

Thanks,

Alex

On 2020-09-22 22:14, Michael Kerrisk (man-pages) wrote:
> On 9/22/20 10:05 PM, Alejandro Colomar wrote:
>> Hi Michael,
>>
>> On 2020-09-22 21:57, Michael Kerrisk (man-pages) wrote:
>>   >
>>   > The previous sentence is rather hard to parse. What (other) part of
>>   > the sentence does "to provide auxiliary information" relate to.
>>   > I suggest splitting the sentence in two and rewording.
>>   >
>>   >
>>   >> +A floating-point control mode is a system variable whose
>>   >
>>   > You suddenly introduce "floating-point control mode" here.
>>   > How does it relate to the preceding sentences? It's not clear.
>>   >
>>   > Going off to look at the standard... I see that actually your
>>   > sentences come pretty much straight from the standard. So, first, I
>>   > think the standard could have been clearer here. Second, and more
>>   > important, (for copyright reasons) we are on shaky ground if we just
>>   > lift whole passages from the standard.  The text does really need to
>>   > be in your own words. Can you come up with something?>
>>   > Alternatively, I guess we could explicitly quote the standard.
>>   > Something like
>>   >
>>   >      POSIX describes this type as follows:
>>   >      .RS
>>   >      .PP
>>   >      [The text]
>>   >      .RE
>>
>> Yes.  A few patches ago I asked about that,
> 
> Sorry -- there was so much mail from you that I missed it!
> 
>> but you didn't answer to that specifically,
>> so I guessed that it was just fair use:
>>
>> https://lore.kernel.org/linux-man/6dc80c25-85bf-925c-49c2-f79865027c0f@gmail.com/T/#mbfdcaf4fe625b4ff7ea90dc7396005fda1283612
> 
> Thanks for the pointer.
> 
>> But I guess explicitly quoting POSIX would be easy and better,
>> as you proposed.
> 
> Yes, I think it''s simplest. (And a patch for 'fenv_t' may be
> a good idea here as well.)
> 
>> I never used that type, so I wouldn't dare to describe it in my own words.
> 
> Okay.
> 
> Thanks,
> 
> Michael
> 
>