Hi Berto, On 1/3/19 12:19 PM, Alberto Garcia wrote: > On Wed 02 Jan 2019 12:09:05 PM CET, Leonid Bloch wrote: >> +print_sizes() { >> + local p=10 >> + while [ ${p} -lt 64 ] >> + do >> + local pad=' ' >> + local n=$((p % 10)) >> + n=$((1 << n)) >> + [ $((n / 100)) -eq 0 ] && pad=' ' >> + [ $((n / 10)) -eq 0 ] && pad=' ' >> + local suff=$((p / 10)) >> + printf "#define S_%u%s%s%20u\n" ${n} "$(size_suffix ${suff})" \ >> + "${pad}" $((1 << p)) >> + p=$((p + 1)) >> + done >> +} > > I have to say that I'm not very convinced of the benefits of replacing a > set of trivial numeric macros with a longer and harder to read shell > script accompanied by changes to the build system.
I think that the benefit is that the script is easily verifiable, whereas if someone would like to verify the table, they will need to generate it themselves. Also, this table is automatically generated anyway, so it only makes sense to generate it during the build. Leonid. > > Berto >