Bug #15043
closed
MinGW intermittent test failure - TestIO#test_select_leak
Description
@normalperson (Eric Wong) Eric,
Thought I'd ping you on this, as the test is very new. The above test is not failing on Travis Ubuntu, nor Appveyor mswin or ruby-loco.
It is intermittently failing on the recently added Ruby Appveyor mingw job. Console output is:
1) Failure:
TestIO#test_select_leak [C:/projects/ruby/test/ruby/test_io.rb:3833]:
pid 1384 exit 1
| #<Thread:0x0000000001f9f8c0@-:11 run> terminated with exception (report_on_exception is true):
| select used to leak on exception (StandardError)
| -: select used to leak on exception (StandardError)
As far as I can tell, there are two main differences between ruby mingw & ruby-loco. First, I'm passing a -j parameter to the test framework, Ruby mingw is passing it to the make command. Secondly, I'm updating the build system (MSYS2/MinGW), Ruby mingw is not. Appveyor can get a few months behind.
I ran the test file a few times locally using runner.rb, both normal & parallel. All tests passed, two skips when run normal, eight skips when run parallel.
Finally, the Appveyor VM's have two procs, I believe Travis' Vm's are four. Build/test times can vary quite a bit on Appveyor.
Thanks, Greg
Updated by normalperson (Eric Wong) over 5 years ago
- Status changed from Open to Closed
Applied in changeset trunk|r64590.
test/ruby/test_io.rb (test_select_leak): use handle_interrupt
Interrupt timing is tricky and it's possible the target
thread is still stopped from the previous loop iteration.
[ruby-core:88732] [Bug #15043]
Updated by normalperson (Eric Wong) over 5 years ago
Greg.mpls@gmail.com wrote:
Bug #15043: MinGW intermittent test failure - TestIO#test_select_leak
https://bugs.ruby-lang.org/issues/15043
Thanks for the report. r64590 ought to fix it
Updated by MSP-Greg (Greg L) over 5 years ago
@normalperson (Eric Wong) Eric,
Thanks for looking at it and also the patch. It passed test-all, but as mentioned, it's intermittent and I can't repo it with ruby-loco... Greg