ngotogenome / gmail.com wrote:
> After r52723, SEGV occurred during Fiddle::TestFunc#test_qsort1 test/fiddle/test_func.rb:83 on Solaris 10 i386 on RubyCI.
> 
http://rubyci.s3.amazonaws.com/unstable10x/ruby-trunk/log/20151123T224815Z.fail.html.gz
> 
> Because this occurred in the CI environment, no other detailed information was available.
> 
> This is not reproduced on sparc Solaris.

Any chance you can log into the i386 VM and reproduce the error
in a standalone script?

In case the qsort implementation on Solaris has global state,
I wouldn't expect our test suite would be running other threads
also calling qsort and messing up the global state.