POSIX Awk says:
The printf statement shall produce output based on a notation similar to the File Format Notation used to describe file formats in this volume of POSIX.1-2008 (see XBD File Format Notation).
And File Format Notation defines %a:
The floating-point number argument representing a floating-point number shall be converted in the style "[-]0xh.hhhhp±d" [...]
However neither Gawk nor Mawk support this:
$ gawk 'BEGIN {printf "%a", 1}'
%a
$ mawk 'BEGIN {printf "%a", 1}'
mawk: run time error: improper conversion(number 1) in printf("%a")
Why is this?
You're looking at the POSIX 2008 SUSv4 (Single Unix Specification) documentation. A lot of software pre-dates this, gawk included. I suspect the gawk implementation is 2001 SUSv2, and has not been updated completely over time.
The Linux (glibc) printf(3) man page alludes to this problem (see the description of %a about half way through, sorry no anchors to link to):
a, A
(C99; not in SUSv2) For a conversion, the double argument is converted to hexadecimal notation (using the letters abcdef) in the style [-]0xh.hhhhp±; [...]
nawk/mawk/gawk don't simply call the underlying libc's printf() or sprintf() verbatim, they more or less reimplement format string processing. For mawk see the do_printf()
function for example. Perl also implements its own format processing, the sprintf man page is more up front about the details.
Things which do support %a/%A are:
printf (i.e. /usr/bin/printf)printf, since bash-2.05 if libc has supportsprintf since perl-5.22.0See the accepted answer here for some additional background: The format specifier %a for printf() in C