Forum: Ruby-core [ruby-trunk - Bug #7213][Open] Namespace regression problem in RSpec from 1.9.2 to 1.9.3

Issue #7213 has been reported by aef (Alexander E. Fischer).

----------------------------------------
Bug #7213: Namespace regression problem in RSpec from 1.9.2 to 1.9.3
https://bugs.ruby-lang.org/issues/7213

Author: aef (Alexander E. Fischer)
Status: Open
Priority: Normal
Assignee:
Category:
Target version:
ruby -v: 1.9.3


This article describes a bug which appeared in 1.9.3: "Including 
namespace module through RSpec config in Ruby 
1.9.3":http://stackoverflow.com/questions/13042298/includ...

Some people believe that it could be related to #3422 which seemed to 
introduce regression but it wasn't sure if this regression was wanted or 
not.

Issue #7213 has been updated by mame (Yusuke Endoh).

Status changed from Open to Assigned
Assignee set to shugo (Shugo Maeda)
Target version set to 2.0.0

If I recall, this is an intentional change since 1.9.3.
Shugo-san, could you check give him a reference to the discussion and 
close this ticket?

--
Yusuke Endoh <mame@tsg.ne.jp>
----------------------------------------
Bug #7213: Namespace regression problem in RSpec from 1.9.2 to 1.9.3
https://bugs.ruby-lang.org/issues/7213#change-32415

Author: aef (Alexander E. Fischer)
Status: Assigned
Priority: Normal
Assignee: shugo (Shugo Maeda)
Category:
Target version: 2.0.0
ruby -v: 1.9.3


This article describes a bug which appeared in 1.9.3: "Including 
namespace module through RSpec config in Ruby 
1.9.3":http://stackoverflow.com/questions/13042298/includ...

Some people believe that it could be related to #3422 which seemed to 
introduce regression but it wasn't sure if this regression was wanted or 
not.

Issue #7213 has been updated by shugo (Shugo Maeda).

Assignee changed from shugo (Shugo Maeda) to mame (Yusuke Endoh)

mame (Yusuke Endoh) wrote:
> If I recall, this is an intentional change since 1.9.3.
> Shugo-san, could you check give him a reference to the discussion and close this 
ticket?

Do you mean Bug #4536?
It seems that foobar_spec.rb fails even if the fix of #4536 (r31215) is 
reverted.

Or do you mean another issue?

I don't know rspec well, but the comment of this ticket stated that this 
issue is related to #3422, and
I don't know anything about #3422.

----------------------------------------
Bug #7213: Namespace regression problem in RSpec from 1.9.2 to 1.9.3
https://bugs.ruby-lang.org/issues/7213#change-32465

Author: aef (Alexander E. Fischer)
Status: Assigned
Priority: Normal
Assignee: mame (Yusuke Endoh)
Category:
Target version: 2.0.0
ruby -v: 1.9.3


This article describes a bug which appeared in 1.9.3: "Including 
namespace module through RSpec config in Ruby 
1.9.3":http://stackoverflow.com/questions/13042298/includ...

Some people believe that it could be related to #3422 which seemed to 
introduce regression but it wasn't sure if this regression was wanted or 
not.

Issue #7213 has been updated by mame (Yusuke Endoh).

Status changed from Assigned to Rejected
Assignee changed from mame (Yusuke Endoh) to shugo (Shugo Maeda)

I meant your answers in these tickets:

https://bugs.ruby-lang.org/issues/6838
https://bugs.ruby-lang.org/issues/5777

As far as I know, #instance_eval with block changed constant lookup 
scope in 1.9.2, but it has stopped since 1.9.3.
The rspec behavior looks affected by this spec change.

I'm closing this ticket, but please tell me if I'm wrong.

Thank you,

--
Yusuke Endoh <mame@tsg.ne.jp>
----------------------------------------
Bug #7213: Namespace regression problem in RSpec from 1.9.2 to 1.9.3
https://bugs.ruby-lang.org/issues/7213#change-32493

Author: aef (Alexander E. Fischer)
Status: Rejected
Priority: Normal
Assignee: shugo (Shugo Maeda)
Category:
Target version: 2.0.0
ruby -v: 1.9.3


This article describes a bug which appeared in 1.9.3: "Including 
namespace module through RSpec config in Ruby 
1.9.3":http://stackoverflow.com/questions/13042298/includ...

Some people believe that it could be related to #3422 which seemed to 
introduce regression but it wasn't sure if this regression was wanted or 
not.

Issue #7213 has been updated by shugo (Shugo Maeda).


mame (Yusuke Endoh) wrote:
> I meant your answers in these tickets:
>
> https://bugs.ruby-lang.org/issues/6838
> https://bugs.ruby-lang.org/issues/5777
>
> As far as I know, #instance_eval with block changed constant lookup scope in 
1.9.2, but it has stopped since 1.9.3.
> The rspec behavior looks affected by this spec change.

In 1.9.2, constant lookup in instance_eval or module_eval had already 
been reverted to the behavior of 1.8,
but there was a bug (#4536) in 1.9.2, and the bug was fixed in 1.9.3.

So I suspected that the fix of #4536 is related to this issue, but 
foobar_spec.rb fails even if the fix is reverted.

Anyway, the current behavior is intended.  The following code 
illustrates why foobar_spec.rb fails in 1.9.3.

  module Foo
   X = 1
  end

  o = Object.new
  o.extend(Foo)
  o.instance_eval {
    p X
  }

This code prints 1 in 1.9.2, but raises a NameError in 1.8 and 1.9.3.
I don't know when this bug of 1.9.2 was fixed.
----------------------------------------
Bug #7213: Namespace regression problem in RSpec from 1.9.2 to 1.9.3
https://bugs.ruby-lang.org/issues/7213#change-32545

Author: aef (Alexander E. Fischer)
Status: Rejected
Priority: Normal
Assignee: shugo (Shugo Maeda)
Category:
Target version: 2.0.0
ruby -v: 1.9.3


This article describes a bug which appeared in 1.9.3: "Including 
namespace module through RSpec config in Ruby 
1.9.3":http://stackoverflow.com/questions/13042298/includ...

Some people believe that it could be related to #3422 which seemed to 
introduce regression but it wasn't sure if this regression was wanted or 
not.