Conversation
Tag a number of specs which have proven to cause intermittent failures. Note that this fixes up the tag for the unstable close_spec which was not being picked up by the tests.
|
Quick update on this: Travis has been really clean lately except when it fails on the tests this pull tags. For instance, the last four runs should have been green: https://travis-ci.org/jruby/jruby/builds/7947845 I'm imagining this hasn't been merged yet to wait and see if these continue coming up (which totally makes sense), so I figured it might be helpful to note they're still an issue. (I also just rebased this branch and re-pushed to trigger a new build. My original push coincided with the tests being in bad shape; my bad) |
|
Ok, merging. I may look at marking these with something other than "fails" since that's the tag we use for stuff known to be unimplemented or buggy; we use e.g. "critical" for failures that are intermittent, bad specs, or environment-unfriendly. |
Tag intermittent spec failures
|
Ah, thanks for the heads up. I should have noticed there's an "unstable" tag, which is clearly more appropriate here. I'll send a follow-up cleaning this up. |
|
Oops...I forgot about "unstable" too and used "critical". Will replace with
|
2 participants
This should improve the stability of the Travis build. Most of these failures are explained in #779 (and marked as such). For the other two:
IO#close on an IO.popen stream sets $?can been seen failing for no good reason here (Note also that this is not a new tag; it simply fixes the tag so that it gets picked up by the build. Copied from the existing tag for 1.8)Process.kill accepts a signal name without the 'SIG' prefixcan be seen failing for no good reason here