### 3.8.2 / 2018-10-09
[Full Changelog](http://github.com/rspec/rspec-expectations/compare/v3.8.1...v3.8.2)
Bug Fixes:
* Change `include` matcher to rely on a `respond_to?(:include?)` check rather than a direct
Hash comparison before calling `to_hash` to convert to a hash. (Jordan Owens, #1073)
* Prevent unexpected call stack jumps from causing an obscure error (`IndexError`), and
replace that error with a proper informative message. (Jon Rowe, #1076)
### 3.8.1 / 2018-08-06
[Full Changelog](http://github.com/rspec/rspec-expectations/compare/v3.8.0...v3.8.1)
Bug Fixes:
* Fix regression in `include` matcher so stopped
`expect(hash.with_indifferent_access).to include(:symbol_key)`
from working. (Eito Katagiri, #1069)
### 3.8.0 / 2018-08-04
[Full Changelog](http://github.com/rspec/rspec-expectations/compare/v3.7.0...v3.8.0)
Enhancements:
* Improve failure message of `change(receiver, :message)` by including the
receiver as `SomeClass#some_message`. (Tomohiro Hashidate, #1005)
* Improve `change` matcher so that it can correctly detect changes in
deeply nested mutable objects (such as arrays-of-hashes-of-arrays).
The improved logic uses the before/after `hash` value to see if the
object has been mutated, rather than shallow duping the object.
(Myron Marston, #1034)
* Improve `include` matcher so that pseudo-hash objects (e.g. objects
that decorate a hash using a `SimpleDelegator` or similar) are treated
as a hash, as long as they implement `to_hash`. (Pablo Brasero, #1012)
* Add `max_formatted_output_length=` to configuration, allowing changing
the length at which we truncate large output strings.
(Sam Phippen #951, Benoit Tigeot #1056)
* Improve error message when passing a matcher that doesn't support block
expectations to a block based `expect`. (@nicktime, #1066)
9deb3c8405