Previously, when performing lazy initialisation of the context,
ApplicationContextRequestMatcher assigned the context field before it
called initialized. The context being non-null is used as the signal
that it’s ok to call a subclass’s matches method. If one thread checks
for a non-null context in between the field being assigned and
initialized being called on another thread, matches will be called
before the subclass is ready.
This commit closes the window for the race condition by only assigning
the context field once the subclass’s initialized method has been
called.
There is a secondary problem in each of the subclasses. Due to the use
of double-checked locking in ApplicationContextRequestMatcher, it’s
possible for a subclass’s matches method to be called by a thread that
has not synchronised on the context lock that’s held when initialized
is called and the delegate field is assigned. This means that the
value assigned to the field may not be visible to that thread.
This commit declares the delegate field of each
ApplicationContextRequestMatcher subclass as volatile to ensure that,
following initialisation, its value is guaranteed to be visible to
all threads.
Closes gh-12380
This commit detects if a `ConnectionNameStrategy` bean exists in the
context and associates it with the auto-configured RabbitMQ's
`ConnectionFactory` when that is the case.
Closes gh-12367
This commit fixes a NPE when the static-locations array of
`ResourceProperties` has to be expanded as the setter is cleaning the
values of the array and is affected by a non-intuitive behaviour of the
binder.
When the binder needs to set an element of an array and the size of the
array isn't large enough, the binder proceeds as follows:
* An array of the required size is created
* The content of the original array is copied over
* The setter of the property is invoked with the new array
* The setter of the property is invoked and the returned array is
mutated to set the requested value
While one would expect the array to contain the requested value when the
setter is invoked, this is not the case. Also, the array might contain
null values if a value at index 8 should be set and the array has a size
of 3.
All in all, `ResourceProperties#appendSlashIfNecessary` has to account
for `null` and an additional round of cleaning has to happen once
binding has completed.
Closes gh-12360
With JDK 8 being the baseline and JDK 7 not being supported anymore we
can get rid of the workaround for a JDK 7 bug in
ProcessBuilder.inheritIO on Windows machines.
Closes gh-12337
Previously, the JDK 10 build would fail as we build the plugin using
Gradle 4.0.x (the lowest version of Gradle that we support) and
Gradle 4.0.x doesn't work with Java 10.
Upgrading to Gradle 4.1, which appears to work with Java 10, was
considered but rejected for now as it introduces the risk that we
inadvertently use an API that's new in 4.1 and break our 4.0 support.
This commit goes for the extreme option and disables building the
Gradle Plugin when building with JDK.
See gh-12028
Andy Wilkinson
Stephane Nicoll
seongwoon.lee
Madhura Bhave
inabajunmr
Cristian Greco
Jon Schneider
dreis2211
Johnny Lim