In Framework 5.2, the Content-Length is now provided by MockMvc. In
earlier versions it is not. This commit updates the tests to tolerate
this difference in behaviour so that the compatibility tests pass
again.
Previously empty parameters in a query string were handled
inconsistently such that they sometimes had a value of "" in a list
and sometimes were represented as an empty list. This inconsistency
lead to the parameter appearing twice in the cURL request snippet.
This commit removes the inconsistency by always using an empty string
to represent an empty query parameter value.
Fixes gh-647
This updates REST Docs to test against Framework's latest 5.1
snapshot. Due to a bug in the behaviour of HttpHeaders' key set [1],
HeaderRemovingOperationPreprocessor has been updated to no longer use
it when removing headers from a request or response prior to it being
documented.
[1] https://github.com/spring-projects/spring-framework/issues/22821
Previously, MockHttpServletRequest allowed its body to be ready
multiple times using its InputStream. As of recent Spring Framework
5.1 snapshots, this is no longer the case.
This commit moves to directly accessing the request's content as a
byte array using getContentAsByteArray() which was introduced in
Framework 5.0.
Closes gh-515
This commit upgrades to Spring Framework 5.0.0.RC1. It also upgrades
the Spring Boot-based samples to use Spring Boot 2.0.0.M1 which
supports Spring Framework 5. Lastly, it upgrades a number of
third-party dependencies to align with Spring Framework 5 and Spring
Boot 2.
Closes gh-386
This commit upgrade to Spring Framework 4.3.6.RELEASE. A side-effect
of this change is that the Content-Type header is now always the first
header when building a request with MockMvc. One test has been updated
to reflect this change in ordering.
Closes gh-342
Previously, when a field was documented it would implicitly document
the whole subsection of the payload identified by that field. This
could lead to users inadvertently failing to document part of the
payload. Arguably, this was a bug as it violated REST Docs' principle
of producing accurate, detail documentation. However, fixing it
requires a breaking change as people may also be relying on this
behaviour. A balance needed to be struck so the fix is being made in
a minor release.
This commit introduces a new subsectionWithPath method which returns a
SubsectionDescriptor; a specialisation of FieldDescriptor. Users
that were intentionally relying on the old behaviour will have to
replace some usage of fieldWithPath with subsectionWithPath instead.
Users who were unintentionally relying on the old behaviour will have
to add some additional descriptors produced using fieldWithPath and
will receive more accurate documentation in return.
Closes gh-274
