Refactor DateTimeFormatterFactory into two distinct classes; a general
purpose factory and a specialized FactoryBean. These changes are
modeled after the existing VelocityEngineFactory and
VelocityEngineFactoryBean classes.
Issue: SPR-9959
Prior to this change, zip archives were named
spring-${version}-${classifier}.zip
e.g.:
- spring-3.2.0.RC1-dist.zip
- spring-3.2.0.RC1-docs.zip
- spring-3.2.0.RC1-schema.zip
This commit updates the Gradle build script to ensure that the fully-
qualified 'spring-framework' name is used consistently, such that names
follow the pattern
spring-framework-${version}-${classifier}.zip
This includes the naming of the root-level directory within the -dist
zip.
Issue: SPR-9954
The DateTimeFormatterFactory introduced in SPR-7121 supports a timeZone
property; however, this property is currently not properly supported.
This commit addresses this issue by ensuring that the timeZone properly
is honored.
Issue: SPR-9953
* SPR-7121:
Support for custom global Joda DateTimeFormatters
Move JRuby dependency below Joda
Support DateTimeFormat annotation without Joda
Corrected date pattern in JavaDocs
Polish whitespace and formatting
Added dateFormatter, timeFormatter and dateTimeFormatter properties
to JodaTimeFormatterRegistrar allowing for custom global formatting.
DateTimeFormatterFactory can be used when configuring with XML.
Issue: SPR-7121
JRuby includes a copy of joda classes with the same package names.
This commit changes the order of the loaded jars to load the original
joda classes first.
Dependency on Joda Time when using the @DateTimeFormat annotation is
now optional. If Joda Time is not present the JDK SimpleDateFormat
will be used to parse and print date patterns. If Joda time is
present it will always be used in preference to SimpleDateFormat.
Issue: SPR-6508
* SPR-9796:
Update cache to support concurrent reads
Cache and late resolve annotations for performance
Polish Property class
Develop ConcurrentReferenceHashMap
Develop a new ConcurrentMap implementation that support both weak and
soft entry references. Intended primarily to be used internally to
support fast concurrent caching.
Issue: SPR-9796
This is more specific exception raised instead of RestClientException
when the raw HTTP status code received from the server is not one of
the HttpStatus enum values.
Issue: SPR-9406
Prior to this commit, calls to ConfigurableEnvironment#addActiveProfile
would cause any active profile values provided via the
"spring.profiles.active" property to be ignored.
Now these two mechanisms can be used in conjunction and work as
expected.
Issue: SPR-9944
Prior to this change,
FrameworkServlet#configureAndRefreshWebApplicationContext called
#postProcessWebApplicationContext(wac) and #applyInitializers(wac)
prior to #refresh, but because servlet-based property source stubs were
not replaced until #refresh, any post-processing or initialization
routines could not benefit from accessing the Environment to retrieve
properties from the ServletContext or ServletConfig.
The workaround to this problem is detailed in SPR-9610 - the user simply
needed to call WebApplicationContextUtils#initServletPropertySources
manually within their ApplicationContextInitializer (or overridden
#postProcessWebApplicationContext method)
This commit ensures that
FrameworkServlet#configureAndRefreshWebApplicationContext calls
WebApplicationContextUtils#initServletPropertySources eagerly, prior to
invoking #postProcessWebApplicationContext and #applyInitializers.
Related Javadoc has also been updated throughout to clarify the behavior
of #initServletPropertySources, when it can be called and what the
effects are, etc.
Note also that a reproduction issue was added to demonstrate the problem
and verify its resolution [1].
[1]: https://github.com/SpringSource/spring-framework-issues/tree/master/SPR-9610
Issue: SPR-9610
1) removed the hamcrest-all dependency requirement and replaced it with
the more focused hamcrest-library dependency
2) added MatcherAssertionErrors as a replacement of
org.hamcrest.MatcherAssert, which in hamcrest 1.1 is only available
through the hamcrest-all dependency (and not in hamcrest-core nor in
the hamcrest embedded in JUnit 4.4 through 4.8)
3) changed the required hamcrest version from 1.1 to 1.3 and made sure
the spring-test-mvc project does not rely on newer hamcrest
functionality without checking if it is available first
Applications that already depend on older versions of hamcrest
(in particular 1.1) via hamcrest-library, hamcrest-all or as part of
junit 4.4 through 4.8 should not be disrupted if they add spring-test
but may wish to exclude the hamcrest-library transitive dependency
from spring-test in order to avoid extra jars in the classpath
Applications that depend on hamcrest 1.3 should not have to do anything
Issue: SPR-9940
This setting was causing failures in the CI build, and was not present
prior to the recent updates to gradle wrapper generation.
This commit also simplifies the gradle wrapper customizations by placing
the logic in a doLast block within the original 'wrapper' task, as
to creating a second wrapper task and extending from the first.
The original commit c9b7b1 ensured the ability to read parameterized
type @RequestBody arguments via GenericHttpMessageConverter (e.g.
application/json and List<String>). However, it also affected the
ability to read @RequestBody arguments that happen are parameterized
but aren't treated as such (e.g. application/x-www-form-urlencoded and
MultiValueMap<String, String>). This commit corrects the issue.
Issue: SPR-9570
For @RequestBody(required=false), it is not sufficient to check if the
InputStream is null. This change reads the first byte of the request
InputStream to determine if the request body is empty or not.
If the InputStream implementation supports mark(int) and reset(), then
we use those. Otherwise we use PushbackInputStream to read and unread
the first byte.
All of this is done only if the required flag of @RequestBody is set
to "false" (default is "true").
Issue: SPR-9942
This commit validates the claims made in SPR-9799.
- Spr9799XmlConfigTests demonstrates that a WAC is not always necessary
when integration testing with XML configuration that uses
<mvc:annotation-driven />.
- Spr9799AnnotationConfigTests demonstrates that a WAC is in fact
necessary when integration testing with a configuration class that
uses @EnableWebMvc.
Issue: SPR-9799
Phillip Webb
Ezequiel Ballesi
Rossen Stoyanchev
Chris Beams
Sam Brannen
Spring Buildmaster
Juergen Hoeller