The build seems ok on my branch, https://travis-ci.org/cammellos/jruby . The test failing seems to be unrelated to the commit, shoud I take a look at it or is ok?

@headius this seems to pass fine and isn't a big change, what do you think including the patch for 9.1 ?

double checked that this wasn't the behaviour in MRI 2.3.1 but the specs seems to match MRI (>=) 2.3.3

@kares Seems appropriate to include it in 9.1.11.

Thanks for the fix and specs!

These specs do not pass on MRI 2.3.3 and 2.3.4:

for ruby in 2.3.3 2.3.4 2.4.0; do chruby $ruby; ruby -v -rbigdecimal -e 'p BigDecimal("-0.2E5556") > -Float::INFINITY'; done
ruby 2.3.3p222 (2016-11-21 revision 56859) [x86_64-linux]
false
ruby 2.3.4p301 (2017-03-30 revision 58214) [x86_64-linux]
false
ruby 2.4.0p0 (2016-12-24 revision 57164) [x86_64-linux]
true

It looks like a MRI bug in 2.3.x.

https://bugs.ruby-lang.org/issues/13674