Stephane Chazelas dixit:
>$ IFS=:
>$ set ${0+:foo:bar}
>$ echo $#
>2
>Expected:
>3
You know, I *loathe* IFS word splitting.
So what’s the rule here? Leading IFS in the variable
after expanding it should be… skipped? Not generate
a field? And leading IFS from the replacement counts?
Or is this about IFS WS vs. IFS NWS?
tglase@tglase:~ $ mksh -c 'IFS=:; set ${0+:foo:bar}; echo $#'
2
tglase@tglase:~ $ ksh93 -c 'IFS=:; set ${0+:foo:bar}; echo $#'
3
tglase@tglase:~ $ mksh -c 'set ${0+ foo bar}; echo $#'
2
tglase@tglase:~ $ ksh93 -c 'set ${0+ foo bar}; echo $#'
2
I don’t get it. And unfortunately, the famous ifs.sh does not
have a testcase for the issue in the first place. (The issue
was IFS-subst-2 and glob-trim-1 in the testsuite.)
Anyway, thanks for reporting…
bye,
//mirabilos
--
“It is inappropriate to require that a time represented as
seconds since the Epoch precisely represent the number of
seconds between the referenced time and the Epoch.”
-- IEEE Std 1003.1b-1993 (POSIX) Section B.2.2.2
