Re: Bugs in printf/sprintf formatted output

[arnold]

Hi.

"Maciej W. Rozycki" <macro@redhat.com> wrote:

>  My suggestion would be making a separate test case (unless GAWK has one 
> already, but your inclusion of the two inputs in `printf-corners.awk' 
> suggests that it doesn't) that covers various inputs of "inf" and "nan" 
> with and without a sign prepended, with and without the `+' and `space' 
> flags, run both in the conventional (or GNU) and in the POSIX mode of 
> GAWK, so that expectations for both modes are covered and things don't 
> accidentally change in the future.  I guess I could have a look into it 
> myself, but I can't commit to a specific timeline.

If you could, I would appreciate it. I am currently refactoring the printf
code, which is a big inline mess. It's taking a lot of my time.

>  NB I have learnt POSIX.1-2024 has been just released to IEEE members and 
> the corresponding HTML version of The Open Group Base Specifications Issue 
> 8 will follow shortly (~July 21st was named).  I guess it will include the 
> `B' and `b' conversions I mentioned before, both for C usage, obviously, 
> and supposedly for AWK as well.

I currently don't plan to add %b or %B. I feel like printing binary
from awk makes little sense, as all awk numbers are floating point. Also,
I doubt that any other awk version is going to add %b/%B anytime soon,
either, and gawk is already enough different from other awks that there's
no reason to add (and have to document!) yet another difference.

Thanks,

Arnold