From: "h.shirosaki (Hiroshi Shirosaki)" Date: 2012-11-24T20:28:28+09:00 Subject: [ruby-core:50037] [ruby-trunk - Bug #7304][Assigned] Random test failures around test_autoclose_true_closed_by_finalizer Issue #7304 has been updated by h.shirosaki (Hiroshi Shirosaki). Status changed from Closed to Assigned % Done changed from 100 to 0 This fix causes segv, which was pointed out at r37831. Thank you, naruse-san. I found rb_ary_delete(ary, obj) is not usable when doing WeakRef finalize because rb_ary_delete() calls rb_equal() against GC'ed WeakRef object. Instead just comparing VALUE by '==' seems good for this case. I'll fix later. ---------------------------------------- Bug #7304: Random test failures around test_autoclose_true_closed_by_finalizer https://bugs.ruby-lang.org/issues/7304#change-33813 Author: luislavena (Luis Lavena) Status: Assigned Priority: High Assignee: h.shirosaki (Hiroshi Shirosaki) Category: core Target version: 2.0.0 ruby -v: ruby 2.0.0dev (2012-11-07 trunk 37538) [i386-mingw32] =begin Hello, Over the past few days I've seen on and off failures on RubyInstaller CI related to (({test_autoclose_true_closed_by_finalizer})): http://ci.rubyinstaller.org/job/ruby-trunk-x86-test-all/265/console 1) Error: test_autoclose_true_closed_by_finalizer(TestIO): NoMethodError: undefined method `close' for 2012-11-07 04:43:41 -0300:WeakRef C:/Users/Worker/Jenkins/workspace/ruby-trunk-x86-build/test/ruby/test_io.rb:1611:in `test_autoclose_true_closed_by_finalizer' This seems to happen when the system is under heavy load (because is running other jobs in parallel). This might be a hint of something not working properly under heavy load, perhaps the GC in effect. I was unable to produce the same failure on x64-mingw32, and haven't tried yet OSX or Linux. Any ideas? =end -- http://bugs.ruby-lang.org/