Bug #1500
closedtest_io runs forever in windows
Description
=begin
Don't know if this is expected, but running
loop {a << IO.pipe; }
runs forever in windows (from test_dup_many).
Thanks!
-=r
=end
Updated by jonco (Jon Coppeard) over 15 years ago
=begin
I am seeing this too.
Ruby version: ruby-1.9.1-p129
Cygwin version: CYGWIN_NT-5.1 1.5.25(0.156/4/2) 2008-06-12 19:34
Windows version: Windows XP Professional Ver 5.1 Build 2600 Service Pack 2
If I remove test_io, then all other tests pass.
Cheers,
Jon
=end
Updated by rogerdpack (Roger Pack) over 15 years ago
=begin
If I remove test_io, then all other tests pass.
They all pass? no failures?
Thanks.
-=r
=end
Updated by yugui (Yuki Sonoda) over 15 years ago
- Assignee set to usa (Usaku NAKAMURA)
- Target version set to 1.9.2
=begin
=end
Updated by usa (Usaku NAKAMURA) over 15 years ago
- Assignee changed from usa (Usaku NAKAMURA) to nobu (Nobuyoshi Nakada)
=begin
cygwin.
=end
Updated by rogerdpack (Roger Pack) over 15 years ago
=begin
Note that this occurs on non cygwin as well.
=end
Updated by hperkins (Hal Perkins) over 15 years ago
=begin
Also observed on Mac OS X 10.5.7 with Ruby 1.9.1-p243. After building with defaults (no options to configure or make), all tests pass except for #226, test_io.rb. That test fails with the message "not finished in 10 seconds [ruby-dev:32566]"
=end
Updated by rogerdpack (Roger Pack) about 15 years ago
=begin
Turns out not to be that
loop {a << IO.pipe; }
runs forever in mingw, but that IO#dup in mingw hangs the process if it is called once, fails, and is called again (same fd).
I'm going to ping the mingw people about it.
Until then could someone please apply the patch to add a timeout, from 373? Otherwise make test-all hangs in mingw.
Thanks!
-r
=end
Updated by rogerdpack (Roger Pack) almost 15 years ago
- Status changed from Open to Third Party's Issue
=begin
Turns out to be a bug in msvcrt.dll [I think].
=end