Bug #3016: Enumerable#slice_before and dup

• Status: Rejected
• Assignee: -
• Target version: 1.9.2
• ruby -v: ruby 1.9.2dev (2010-03-03 trunk 26805) [x86_64-darwin10.2.0]

[marcandre (Marc-Andre Lafortune)]

=begin
When writing the specs for slice_before, I was surprised that it doesn't call #dup.

class X
def dup
raise "Hi!"
end
end

[:foo].slice_before(X.new){}.to_a

==> [[:foo]], I expected it to raise

The implementation calls directly rb_obj_dup.

It's a minor issue, but would it not be best to call dup on the argument? Most special needs for #dup can be addressed by initialize_dup, but maybe there are some reasonable cases where some user classes will redefine #dup?

The same question is valid for Enumerable#chunk.

diff --git a/enum.c b/enum.c
index b69d8c9..e6e6adb 100644
--- a/enum.c
+++ b/enum.c
@@ -14,7 +14,7 @@
#include "node.h"

VALUE rb_mEnumerable;
-static ID id_each, id_eqq, id_cmp, id_next, id_size;
+static ID id_each, id_eqq, id_cmp, id_next, id_size, id_dup;

static VALUE
enum_values_pack(int argc, VALUE *argv)
@@ -2189,7 +2189,7 @@ chunk_i(VALUE yielder, VALUE enumerator, int argc, VALUE *argv)
arg.yielder = yielder;

  if (!NIL_P(arg.state))

•    arg.state = rb_obj_dup(arg.state);
  

•    arg.state = rb_funcall(arg.state, id_dup, 0, 0);
  

  rb_block_call(enumerable, id_each, 0, 0, chunk_ii, (VALUE)&arg);
  if (!NIL_P(arg.prev_elts))
  @@ -2364,7 +2364,7 @@ slicebefore_i(VALUE yielder, VALUE enumerator, int argc, VALUE *argv)
  arg.yielder = yielder;

  if (!NIL_P(arg.state))

•    arg.state = rb_obj_dup(arg.state);
  

•    arg.state = rb_funcall(arg.state, id_dup, 0, 0);
  

  rb_block_call(enumerable, id_each, 0, 0, slicebefore_ii, (VALUE)&arg);
  if (!NIL_P(arg.prev_elts))
  @@ -2606,5 +2606,6 @@ Init_Enumerable(void)
  id_cmp = rb_intern("<=>");
  id_next = rb_intern("next");
  id_size = rb_intern("size");

• id_dup = rb_intern("dup");
  }
  =end

[matz (Yukihiro Matsumoto)]

=begin
Hi,

In message "Re: [ruby-core:29026] [Bug #3016] Enumerable#slice_before and dup"
on Fri, 26 Mar 2010 14:02:05 +0900, Marc-Andre Lafortune [e-mail] writes:

|When writing the specs for slice_before, I was surprised that it doesn't call #dup.

MRI often bypasses method calls using direct C function call, mostly
for performance reason. I am afraid that it's too late to fix all
"surprises". I say it's implementation dependent behavior. I think
it is over specification to check.

						matz.

=end

[mame (Yusuke Endoh)]

• Status changed from Open to Rejected

=begin
Hi,

This ticket is rejected by matz.
So closed.

--
Yusuke Endoh [e-mail]
=end