Issue #1125 has been updated by Yusuke Endoh.

Assignee changed from Yukihiro Matsumoto to Yusuke Endoh

Hello,

I'll commit the following patch unless there is objection:

diff --git a/compile.c b/compile.c
index 32bda52..9ea2a3c 100644
--- a/compile.c
+++ b/compile.c
@@ -4644,6 +4644,8 @@ iseq_compile_each(rb_iseq_t *iseq, LINK_ANCHOR *ret, NODE * node, int poped)
       case NODE_SPLAT:{
 	COMPILE(ret, "splat", node->nd_head);
 	ADD_INSN1(ret, nd_line(node), splatarray, Qfalse);
+	ADD_INSN1(ret, nd_line(node), newarray, INT2FIX(0));
+	ADD_INSN(ret, nd_line(node), concatarray);
 
 	if (poped) {
 	    ADD_INSN(ret, nd_line(node), pop);

-- 
Yusuke Endoh <mame / tsg.ne.jp>
----------------------------------------
Feature #1125: [*x] (array consisting only of a splat) does not necessarily return a new array
https://bugs.ruby-lang.org/issues/1125

Author: Daniel Luz
Status: Assigned
Priority: Normal
Assignee: Yusuke Endoh
Category: 
Target version: 2.0.0
ruby -v: ruby 1.8.7 (2008-08-11 patchlevel 72) [x86_64-linux]


=begin
 For [*x], these are basically the possible outcomes:
 
 1. if x is an Array, returns it unmodified.
 2. elsif x responds to to_ary (to_a on 1.9.1), invokes that method and returns its result unmodified.
 3. else, returns a new array with x as its only element.
 
 Cases #1 and #2 IMO violate the POLS, as I expected that an array literal would always return a new array. The practical consequence here is that I expected I'd be free to modify it without side effects. (For comparison, "#{x}" always returns a new string)
 
 Simple test case:
   x = [1, 2, 3]
   [*x] << 4
   p x # => [1, 2, 3, 4]
 
 
 Thus, I propose ensuring these two cases always return new arrays.
 One possible solution would be dup'ing the resulting array (I guess that'd have a rather low cost; the third case would result in an unnecessary dup, but at least it's just a single-item array). Another one would be to dumb down the interpreter, making it create a zero-length array and then concat the result of the splat to it.
=end



-- 
http://bugs.ruby-lang.org/