[ruby-core:72181] [Ruby trunk - Bug #11811] Chaining lazy enumerators causes duplicate ouput

Issue #11811 has been updated by Shugo Maeda.


Shugo Maeda wrote:
> Nobuyoshi Nakada wrote:
> > I'm afraid that it may cause GC mark miss of dynamic symbols.
> 
> Ah, I see.
> 
> How about to apply the optimization only to static symbols?

Or the following patch also seems to fix the bug:

```
diff --git a/vm.c b/vm.c
index 087cba7..a03e068 100644
--- a/vm.c
+++ b/vm.c
@@ -567,6 +567,10 @@ vm_make_proc_from_block(rb_thread_t *th, rb_block_t *block, VALUE *procptr)
 	*procptr = block->proc = rb_vm_make_proc(th, block, rb_cProc);
 	return TRUE;
     }
+    else if (SYMBOL_P(block->proc)) {
+	*procptr = rb_sym_to_proc(block->proc);
+	return TRUE;
+    }
     else {
 	*procptr = block->proc;
 	return FALSE;
```


----------------------------------------
Bug #11811: Chaining lazy enumerators causes duplicate ouput
https://bugs.ruby-lang.org/issues/11811#change-55595

* Author: Chris Beer
* Status: Open
* Priority: Normal
* Assignee: Nobuyoshi Nakada
* ruby -v: ruby 2.3.0preview2 (2015-12-11 trunk 53028) [x86_64-darwin15]
* Backport: 2.0.0: UNKNOWN, 2.1: UNKNOWN, 2.2: UNKNOWN
----------------------------------------
In Ruby 2.3.0-preview2, I'm seeing a change in behavior using lazy enumerators with select/reject and the & operator:

irb(main):037:0> %w(1 2 3).lazy.reject(&:empty?).each { |x| puts x }
1
1
2
2
3
3

Note that the output is doubled. However, if I don't use the & shorthand, the output is as expected:


irb(main):038:0> %w(1 2 3).lazy.reject { |x| x.empty? }.each { |x| puts x }
1
2
3
=> nil


And in Ruby 2.2.3, both variants produced the same result:

irb(main):001:0> %w(1 2 3).lazy.reject(&:empty?).each { |x| puts x }
1
2
3
=> nil


---Files--------------------------------
vm_caller_setup_arg_block.diff (1.03 KB)


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