[BUG] SHWORDSPLIT: unset IFS: $* and $@ don't split

[BUG] SHWORDSPLIT: unset IFS: $* and $@ don't split

[Martijn Dekker]

In SHWORDSPLIT mode, if IFS is unset (meaning we expect default field
splitting),  unquoted $* and $@ (including in substitutions like
${1+$@}) or ${var-$*} do not perform the expected default field
splitting after generating fields from the positional parameters.

#! Src/zsh -f
setopt shwordsplit
set -- ab 'cd ef' gh
unset -v IFS
printf '[%s]\n' $* $@

Actual output:
[ab]
[cd ef]
[gh]
[ab]
[cd ef]
[gh]

Expected output:
[ab]
[cd]
[ef]
[gh]
[ab]
[cd]
[ef]
[gh]

Ref.:
http://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html#tag_18_05_02
For both $@ and $*, their behaviour is only different if they are
quoted. Unquoted, they behave identically:
| Expands to the positional parameters, starting from one, initially
| producing one field for each positional parameter that is set. When
| the expansion occurs in a context where field splitting will be
| performed, any empty fields may be discarded and each of the
| non-empty fields shall be further split as described in Field
| Splitting.

Note that it says "any empty fields *may* be discarded" (not "shall" be
discarded). Zsh discards them, as does nearly every other shell; yash as
of 2.44 is the only one I know of that doesn't. The description at
http://austingroupbugs.net/view.php?id=888 indicates the behaviour may
become mandatory in the future.

Thanks,

- M.

Re: [BUG] SHWORDSPLIT: unset IFS: $* and $@ don't split

[Bart Schaefer]

On Feb 19,  6:45am, Martijn Dekker wrote:
} Subject: [BUG] SHWORDSPLIT: unset IFS: $* and $@ don't split
}
} In SHWORDSPLIT mode, if IFS is unset (meaning we expect default field
} splitting),  unquoted $* and $@ (including in substitutions like
} ${1+$@}) or ${var-$*} do not perform the expected default field
} splitting after generating fields from the positional parameters.

This one turns out to be more tractible.

diff --git a/Src/subst.c b/Src/subst.c
index 670f3f0..1c2397c 100644
--- a/Src/subst.c
+++ b/Src/subst.c
@@ -3476,7 +3476,8 @@ paramsubst(LinkList l, LinkNode n, char **str, int qt, int pf_flags,
 		val = sepjoin(aval, sep, 1);
 		isarr = 0;
 		ms_flags = 0;
-	    } else if (force_split && (spsep || nojoin == 2)) {
+	    } else if (force_split &&
+		       (spsep || nojoin == 2 || (!ifs && isarr < 0))) {
 		/* Hack to simulate splitting individual elements:
 		 * forced joining as previously determined, or
 		 * join on what we later use to forcibly split