New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
libxml 2.8.0 test suite hangs on ruby 2.2 #106
Comments
Hmm, have you done any debugging into which test is hanging? |
I've just retested with ruby 2.2.1p85 (2015-02-26 revision 49769) [x86_64-linux] and have some more information. I'm observing the hang in tc_node_edit.rb in the test_merge method, while reading the 500 documents. Adding a counter shows that I'm getting from 0 to 160 and then the busy loop happens. ruby just sits using 100% memory with apparently constant memory. Aha, and the 160 is not constant, I'm seeing different values here. I've even had a single run where it passed through test_merge, and then exhibited the same behavior later. In fact, commenting out tc_node_edit.rb in test_suite.rb does not solve the issue because later tests then seem to trigger it. My best guess: a bug in ruby's GC is being triggered somehow. The ruby process doesn't listen to SIGTERM. I've added GC.start inside the 500.times loop and with that in place the test_merge test passes consistently every time. |
Hi! I'm observing a similar behavior on Debian with ruby2.2 (2.2.2p95). |
I have the same issue with 2.2 on Ubuntu in test_memory in test/tc_xpath.rb. |
Yes, I sometimes see this too. It does seem to have to do with garbage collection, but after playing with it for a bit nothing obvious stands out yet. |
Finally tracked this down! |
Thanks! I no longer see test suite hangs with ruby22 and libxml 2.9.0. |
Great - thanks for checking. |
With ruby 2.2.0p0 (2014-12-25 revision 49005) [x86_64-linux] the libxml 2.8.0 test suite hangs indefinitely in a busy loop. The test suite does pass with other ruby versions, e.g. ruby 2.1.5p273 (2014-11-13 revision 48405) [x86_64-linux]
The text was updated successfully, but these errors were encountered: