Test in Ruby 2.7#7386
Conversation
|
cc: @haberman |
|
Could you sync? |
188adee to
9e48c9f
Compare
|
/usr/local/rvm/rubies/ruby-2.7.0/lib/ruby/2.7.0/rubygems/core_ext/kernel_require.rb:72: [BUG] Segmentation fault at 0x0000000102794458 |
|
The segfault occurs before even loading |
|
@qnighy thank you! I was not looking forward to debugging that one... |
|
Is there a way we can disable the test(s) that are triggering this Ruby bug, until the bug is fixed upstream? I am planning to start a release very soon and was hoping Ruby 2.7 gems would be part of it. |
|
Is there a more recent ruby version with the fix?
…On Wed, Apr 22, 2020 at 12:20 PM Joshua Haberman ***@***.***> wrote:
Is there a way we can disable the code that is triggering this Ruby bug,
until the bug is fixed upstream?
I am planning to start a release very soon and was hoping Ruby 2.7 gems
would be part of it.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#7386 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABHUPZKNNIFFLDYMKJMY753RN47PNANCNFSM4MJBUUJQ>
.
|
|
Oh, just found it. I think we can disable the test. If possible, disable
only for ruby 2.7
…On Wed, Apr 22, 2020 at 12:41 PM Joshua Haberman ***@***.***> wrote:
@TeBoring <https://github.com/TeBoring> The issue was only reported
upstream yesterday by @qnighy <https://github.com/qnighy> . I don't think
the Ruby developers had heard about the bug before then. So I expect there
is no released fix yet.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#7386 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABHUPZKEVT7OU6DIQCWI653RN5B7XANCNFSM4MJBUUJQ>
.
|
|
It appears that it's the code loading itself that is triggering the bug, not a test. So maybe we need to disable |
I added the following commit (d95f630). How about this? diff --git a/ruby/tests/gc_test.rb b/ruby/tests/gc_test.rb
index 55b96289e..6ef4e2e30 100755
--- a/ruby/tests/gc_test.rb
+++ b/ruby/tests/gc_test.rb
@@ -4,7 +4,9 @@
$LOAD_PATH.unshift(File.expand_path(File.dirname(__FILE__)))
old_gc = GC.stress
-GC.stress = 0x01 | 0x04
+# Ruby 2.7.0 - 2.7.1 has a GC bug in its parser, so turn off stress for now
+# See https://bugs.ruby-lang.org/issues/16807
+GC.stress = 0x01 | 0x04 unless RUBY_VERSION.match?(/^2\.7\./)
require 'generated_code_pb'
require 'generated_code_proto2_pb'
GC.stress = old_gc |
|
Looks like we have one failure now: |
|
Fixed in 973df41 |
Split from #7027. This PR focuses on testing on Ruby 2.7, while #7027 focuses on building artifacts for Ruby 2.7.
I guess someone needs to run
kokoro/linux/dockerfile/push_testing_images.shbefore running CI, because I modified a dockerfile inkokoro/linux/dockerfile/test?