CI: Enable jruby-head, don't allow jruby releases to fail#145
Conversation
|
This will fail on the "jruby" matrix due to 10.0.1.0 having the bug reported in jruby/jruby#8920, but ideally should only show the two issues I mentioned in #72 (comment) when running against "jruby-head". I would appreciate some pointers to help me figure out why those failures happen. |
|
jruby-head has been updated to include the fix for jruby/jruby#8920. I'd like to see if those same two failures occur in CI... can someone approve the workflow? |
|
I just approved CI to run (FYI I'm on vacation and can't be of much help for a few weeks) |
|
Good news! The jruby-head build runs as it does for me locally, with only two failure I will look into. All fixes on head should be in a JRuby release within the next couple of weeks. I will update again once I have fixed the remaining two failures. |
|
Actually I'm a bit confused. The failing code appears to be attempting to load |
|
Oh I get it... it's supposed to fail, but for some reason the failure is not bubbling out to the spec properly. |
This was changed in CRuby in 2010. See ruby/ruby@208a51f This fixes the remaining two failures running mustermann specs. See sinatra/mustermann#145
|
Aha! It is a difference of error message due to a change we apparently missed way back in 2010. I've pushed a PR for JRuby to fix this. With the patch in place, mustermann's specs are green! |
|
I have merged the last pull request to JRuby and initiated a head build here: This should be ready shortly, and then the JRuby head build here should be green. We could remove the jruby release job for now if you prefer, but we should be releasing JRuby 10.0.2.0 this week. |
|
I'll merge this when JRuby 10.0.2.0 is out! |
|
@dentarg Could you rerun the latest build so I can confirm head is green please? Thank you! |
|
@headius can you rebase and push? that should trigger builds again I think Despite |
|
Can do! |
|
Only JRuby release build that's failing, as expected! |
|
I saw that JRuby 10.0.2.0 is out https://www.jruby.org/2025/08/07/jruby-10-0-2-0.html 🎉 Now we wait for rbenv/ruby-build#2558 to be merged and for the automation in https://github.com/ruby/ruby-builder to pick it up and make it available in https://github.com/ruby/setup-ruby |
|
It's a long tail but we are almost there! |
|
It appears JRuby 10.0.2.0 has propagated through ruby-build and ruby-builder! I'll rebase one more time to ensure it's working properly. |
In order to help ensure JRuby does not break mustermann again, remove the allow-failure flag and add jruby-head with sinatra head for more coverage. The issue reported in jruby/jruby#8920 wouldn't have been released if it had been caught by jruby-head testing in mustermann, and having allow-failure means bugs introduced by JRuby may never be caught by mustermann CI. I will push additional fixes to get the remaining failure green.
dentarg
changed the title
2 participants
In jruby/jruby#8922 a bug was reported in JRuby 10.0.1.0 that might have been caught before release had mustermann's CI been running
jruby-headbuilds without enablingallow-failure. This PR will modify mustermann CI to run bothjrubyandjruby-head, the latter against sinatra head, without allowing failures to be silently ignored (because we'll fix them).