Use DataBufferUtils.compose instead of writeAggregator to combine
multiple data buffers into one, as the write aggregator would not work
when the initial data buffer did not have enough capacity to contain
all subsequent buffers.
Removed writeAggregator, as it is no longer needed.
Issue: SPR-16365
This allows us to confirm 0.7.3 fixes and run tests against the latest
ahead of an upgrade to Reactor Netty 0.8 snapshots.
The fix for #231 and #240 are confirmed but unfortunately test case for
With this commit, the default content-type defined by the
view (usually "text/html;charset=UTF-8" defined in AbstractView)
is used if any, when none is defined in the response headers.
Issue: SPR-16247
If the response is set and we can't change the status through
ServerHttpResponse any more, allow the error signal to propagate and
let the individual server adapters handle it. Ultimately that should
result in closing the connection.
On Servlet containers, we check one last time if the response is
committed (we may not have filled the buffer). If not then save
the exception as a request attribute, dispatch, and re-throw it on the
container thread.
On Undertow access the connection and close it.
On Netty just let the error through to Reactor Netty.
Issue: SPR-16051
Access-Control-Allow-Credentials CORS header, used to
allow cookies with CORS requests, is not set to true
anymore by default when enabling CORS with
@CrossOrigin or global CORS configuration in order to
provide a more secured default CORS configuration.
The related allowCredentials property now requires to
be set to true explicitly in order to support cookies
with CORS requests.
Issue: SPR-16130
This commit adds
AbstractMessageReaderArgumentResolver#readBody and
AbstractMessageWriterResultHandler#writeBody variants
which allow to pass the actual MethodParameter in order
to perform proper annotation-based hint resolution with
nested generics, for example with HttpEntity.
Issue: SPR-16098
Practically no changes to RxNetty for a year and efforts underway to
rebuild 0.6.x based on a current Reactor Netty base.
Aside from the extra time to run integration tests having two
Netty-based servers can also cause false alarms such as ByteBuf leaks
related to RxNetty.
FromHttpMessageWriter and MultipartHttpMessageWriter both support
MultiValueMap except the former supports String values only. This
presents an issue since either full generic type information must be
provided, which is cumbersome on the client side, or if left out there
is no good way to order the writers to make a proper decision.
This commit:
- refines the canWrite behavior of to not a
accept MultiValueMap without proper generic information unless the
MediaType is explicitly set providing a strong hint.
- modifies MultipartHttpMessageWriter to be configured with a
FormHttpMessageWriter so it can write both plan and multipart data with
the ability to properly differentiate based on actual map values.
Issue: SPR-16131
This commit exposes the ClientRequest's URL and HttpMethod fields via a
setter, so that they can be changed more easily in a request that was
created via ClientRequest.from(ClientRequest).
Issue: SPR-16093
Before this commit, when adding filters to a builder obtained via
`WebClient.mutate()`, the filters were added both to the original client
as well as the mutated builder. This commit fixes that.
Issue: SPR-16059
Prior to this commit, the `ResourceWebHandler` would itself handle the
response with an HTTP 404 in many cases, including a missing static
resource.
This does not give a chance to `WebExceptionHandler` instances to handle
that error and, for example, display an error page.
See spring-projects/spring-boot#8625
Issue: SPR-16023
Concrete server and client, reactive request and response
implementations should not have to be accessed outside their package.
They could be seen as private to their HttpHandler adapters and
ClientHttpConnector's respectively.
The one exception, WebSocket upgrades in spring-webflux, is an internal
framework use case, accommodated via downcast + accessors in the
abstract base classes.
Prior to this commit, the `WebClient` always throws a `ClassCastException`
when an error occurs in `bodyToMono(ParameterizedTypeReference)``, and
not the expected exception, as set up by `onStatus`
Issue: SPR-16025
Prior to this commit, asking for a `Void` type using any of the
`ClientResponse#bodyTo*` methods would immediately return an empty
`Publisher` without consuming the response body.
Not doing so can lead to HTTP connection pool inconsistencies and/or
memory leaks, since:
* a connection that still has a response body being written to it cannot
be properly recycled in the connection pool
* incoming `DataBuffer` might not be released
This commit detects when `Void` types are asked as body types and in
those cases does the following:
1. Subscribe to the response body `Publisher` to allow the connection to
be returned to the connection pool
2. `cancel()` the body `Publisher` if the response body is not empty; in
that case, we choose to close the connection vs. consume the whole
response body
Those changes imply that `ClientHttpResponse` and other related
contracts don't need a `close()` method anymore.
Issue: SPR-16018
Prior to this issue, SPR-15920 added this new `close()` method which was
supposed to be called to clean resources after response processing.
This commit changes the meaning of that method: calling `close()` will
close the underlying HTTP connection. This has to be called if the
response body is not consumed by the application, since at that point
the underlying connection might be in an inconsistent state if shared in
a connection pool.
Issue: SPR-15993
