Enumerator#any? is wrong for sequence of empty arrays

[buehmann]

Environment

$ cat jruby_bug.rb 
p [[]].any?
p [[]].to_enum.any?

Expected Behavior

$ ruby -v                      
jruby 9.2.8.0 (2.5.3) 2019-08-12 a1ac7ff OpenJDK 64-Bit Server VM 11.0.5-ea+10-post-Ubuntu-0ubuntu1 on 11.0.5-ea+10-post-Ubuntu-0ubuntu1 +jit [linux-x86_64]
$ ruby jruby_bug.rb 2>/dev/null
true
true

(same for MRI)

Actual Behavior

$ ruby -v                      
jruby 9.2.9.0 (2.5.7) 2019-10-30 458ad3e OpenJDK 64-Bit Server VM 11.0.5-ea+10-post-Ubuntu-0ubuntu1 on 11.0.5-ea+10-post-Ubuntu-0ubuntu1 +jit [linux-x86_64]
$ ruby jruby_bug.rb 2>/dev/null
true
false

Discovered through rubocop/rubocop#7489

[ahorek]

bisected to this change d16918b

here's a different example that better explains the behavior

actual

 [[]].to_enum.take(1)
=> [nil]

expected

 [[]].to_enum.take(1)
=> [[]]

cc @kares

[kares]

oh nice, so we got a regression - seems like the test coverage isn't perfect.

was basically trying to increase compatibility (there's some more commits involved with Enumerator work in 9.2.9 than the one @ahorek bisected for us) and fix some long standing bugs.

hopefully we can fix this (pathological) case without regressing other stuff we started passing in 9.2.9

[koic]

Hi @kares. This issue doesn't seem to be resolved in JRuby 9.2.10.0 and 9.2.11.0.

% cat jruby_bug.rb
p [[]].any?
p [[]].to_enum.any?

JRuby 9.2.8.0

% ruby -v
jruby 9.2.8.0 (2.5.3) 2019-08-12 a1ac7ff Java HotSpot(TM) 64-Bit Server VM 25.5-b02 on 1.8.0_05-b13 +jit [darwin-x86_64]
% ruby jruby_bug.rb
true
true

JRuby 9.2.10.0

% ruby -v
jruby 9.2.10.0 (2.5.7) 2020-02-18 fffffff Java HotSpot(TM) 64-Bit Server VM 25.5-b02 on 1.8.0_05-b13 +jit [darwin-x86_64]
% ruby jruby_bug.rb
true
false

JRuby 9.2.11.0

% ruby -v
jruby 9.2.11.0 (2.5.7) 2020-03-02 612d7a05a6 Java HotSpot(TM) 64-Bit Server VM 25.5-b02 on 1.8.0_05-b13 +jit [darwin-x86_64]
% ruby jruby_bug.rb
true
false

[headius]

Retargeted to 9.2.12. See #6285.