Previously, a relative PID folder was not handled correctly when
running stop, status, or force_reload. This meant that a service
could be started when configured to use a relative pid file, but
then could not be stopped.
The PID folder should be treated as relative to the service's jar
file. This commit updates stop, status, and force_reload to push the
jar file's directory so that this is now the case for those three
commands.
Closes gh-7092
To be compatible with Gradle's plugin portal, plugins must have an
ID that uses a reverse domain name. This means that spring-boot is
not compatible.
This commit introduces a new ID, org.springframework.boot, and
deprecates the old ID.
Closes gh-6997
This is an alternative to the fix made in 3b52909 which removed the
chown call entirely.
Prior to 3b52909, the ownership of $PID_FOLDER was always changed even
when its value was /var/run. This was problematic as it could prevent
other services from creating their pid folder or file.
When a sub-folder is used, changing its ownership so that it’s owned by
the user that will run the app is desirable as it limits access to the
folder. Rather than removing the chown call entirely, this commit
ensures that it only happens when a sub-folder is being used to hold the
pid file.
Closes gh-6532
Update the launch script so that it no longer changes ownership of the
PID_FOLDER.
Commit b24e736cfe had changed the chown
line from:
chown "$run_user" "$PID_FOLDER/${identity}"
to:
chown "$run_user" "$PID_FOLDER"
This meant that it was possible for the launch script to change
ownership of `/var/run` and prevent later processes from writing to
the folder.
Since PID_FOLDER is created before the chown statement, and that
the `checkPermissions` function runs to ensure that the PID file can
be written, it appears that the chown is not even required.
Fixes gh-6532
This commit improves the run goal to automatically fork the process when
devtools is present and log a warning when fork has been disabled via
configuration since devtools will not work on a non-forked process.
We don't want devtools to kick in for integration tests so the logic has
been placed in `RunMojo` requiring a couple of protected methods to
override.
Closes gh-5137
So far, one has to set the "fork" value to both the start and stop
goals. Since they have the same name, sharing them in a global
configuration element does the trick. However, the plugin also supports
auto-detection of the fork value according to other parameters:
typically if an agent or jvm arguments are set, forking will be
automatically enabled. This is a problem since the stop goal is not aware
of that.
This commit transmits the value in a property attached to the
`MavenProject`. That way, the stop goal can retrieve that value and
apply the same defaults. This has the side effect that specifying the
fork value isn't necessary anymore.
Closes gh-6747
Previously, configuring the build-info goal in a pom would result in
Eclipse reporting an error for the pom as it didn’t know if/when to
execute the build-info goal.
This commit adds lifecycle mapping metadata so that the goal is executed
on incremental builds. This ensures that the contents of the generated
file are kept up-to-date, reflecting the latest build time, etc.
Closes gh-6723
Previously, RandomAccessDataFile used a semaphore and acquired it
interruptibly. This meant that an interrupted thread was unable to
access the file. Notably, this would prevent LaunchedURLClassLoader from
loading classes or resources on an interrupted thread.
The previous commit (937f857) updates RandomAccessDataFile to acquire
the semaphore uninterruptibly. This commit adds a test to
LaunchedURLClassLoader to verify that it can now load a resource from
an interrupted thread.
Closes gh-6683
Previously, if mvn spring-boot:run with a forked JVM was killed with
CTRL+C, the run would be considered unsuccessful. This commits updates
the run mojo to consider a forked JVM that exists with 130 (the exit
code produced when exiting due to SIGINT which is what CTRL+C sends) to
be successful.
Closes gh-6498
The unpack location is (largely) an implementation detail and the
Repackage Mojo was the only place where it was explicity documented.
Rather than updating the outdated location, this commit removes it
entirely to avoid encouraging people to rely on the location.
Closes gh-6624
META-INF/INDEX.LIST files are pointless in an executable jar and
moving application classes from the root of the jar to
BOOT-INF/classes breaks the index, resulting in an
InvalidJarIndexException being thrown.
This commit updates the Repackager to automatically remove a
META-INF/INDEX.LIST file from a jar file that is being repackaged.
Closes gh-6601
Where possible, code that previously synchronized on this (or on the
class in the case of static methods) has been updated to use an
internal monitor object instead. This allows the locking model that's
employed to be an implementation detail rather than part of the
class's API.
Classes that override a synchronized method continue to declare
the overriding method as synchronized. This ensures that locking
is consistent across the superclass and its subclass.
Closes gh-6262
Previously, if a property name had successive capital letters, the
generated meta-data would clean it in such a way it is defined as a
regular word. For instance a `myFOO` property would be written as
`my-foo` in the meta-data.
It turns out this decision is wrong as the binder has no way to compute
back the name of the property and therefore `my-foo` wouldn't bind to
`setMyFOO` as it should.
This commit updates the meta-data name generation algorithm to properly
identify such cases: `myFOO` now translates to `my-f-o-o`. While the
generated name is a bit ugly, it now provides a consistent binding
experience.
Closes gh-5330
This commit improves the performance of JarURLConnection. There are two
main changes:
Firstly, the way in which the spec is determined has been changed so
that it’s no longer necessary to create an absolute file. Instead,
the JarFile’s pathFromRoot is used to extract the spec from the URL
relative to the JarFile.
Secondly, the number of temporary Objects that are created has been
reduced, for example by tracking an index as we process a String
rather than creating a new substring for each iteration.
See gh-6215
JarURLConnection is very performance sensitive. The change in 3772d9f
meant that every JarURLConnection would create a FilePermission,
irrespective of whether it was actually used.
This commit updates JarURLConnection to create its FilePermission
lazily. When there is no security manager a permission will no longer
be created at all.
Closes gh-5411
See gh-6215
Previously, JarURLConnection assumed that that URL with which it was
created would contain the absolute path of the underlying jar file.
This meant that when it was created with a relative URL, it could fail
to find an entry or throw a StringIndexOutOfBoundsException.
This commit updates the logic for normalizing the input URL so that
both absolute and relative URLs are supported.
Closes gh-6109
This commit detects case where the same set of keys are exposed several
times and prevents the compilation to complete. Previously, duplicate
keys were silently added to the meta-data.
Closes gh-5939
In fact the folder was already created if the app is running as
a different user, but not if running as the current user, so it
was just a question of moving one line out of an if block.
Fixes gh-5986
Update TypeUtils to guard against the use of older Java versions.
Both `Collection` and `Map` type lookups now fallback to generic free
versions of the classes.
Prior to this commit using `xmlbeans-maven-plugin` in combination with
Spring Boot's annotation processor could result in
`IllegalArgumentException: Incorrect number of type arguments`.
Fixes gh-6122
Previously, Maven repositories definition was specified in a profile that
is active by default. It means that as soon as any profile is enabled by
the user, said profile is no longer enabled. This has the nasty
consequences of having copy/paste in several places to make sure our own
profiles still have the proper repositories definition.
This commit creates a single "repositories" profile that is always active
unless a given property is explicitely specified. This allows to remove
the duplication and make things more consistent.
Some Gradle-specific repositories were also hard-coded in two modules
without any profile at all, meaning they were polluting the build of
anybody using it. While the impacted modules are gradle specific, that
repository has been shared in the new "repositories" profile as well.
Closes gh-6031
Previously, JarURLConnection didn't override getPermission(). This
meant that it required all permissions. This was at odds with the
Oracle JVM's concrete sun.net.www.protocol.jar.JarURLConnection which
overrides getPermission to return a FilePermission with the read
action for the path of the underlying jar.
This commit updates our JarURLConnection to align its behaviour with
sun.net.www.protocol.jar.JarURLConnection.
Closes gh-5411
In Maven land, provided and system-scope dependencies are very similar,
the latter being an special kind that allows you to specify the path to
the artifact rather than using the repository to locate it.
Prior to this commit, the repackage goal of the maven plugin was
inconsistent as it would repackage provided-scope dependencies but would
ignore the system-scoped ones.
This commit adds an extra boolean flag, `includeSystemScope` to control
this behaviour. For backward compatibility reasons, its default value is
`false`.
Closes gh-2224
Previously, arguments passed to the script were handled in one way
if a service was being started using start-stop-daemon and in another
way if start-stop-daemon wasn’t available or the application is being
launched in run mode. This meant that quoted arguments were only
handled correctly when the application was being started using
start-stop-daemon.
This commit updates the launch script so that argument handling is
the same across all three different way that the application can be
launched.
Closes gh-5942
Using a separate thread to call the application's main method is
unnecessary – the context class loader of the current thread can be
updated instead – and makes exception and exit code handling more
complicated than it needs to be.
This commit updates the Launcher so that it calls the main method
runner using the current (main) thread. As a result, any exception
that's thrown will be caught by the JVM and result in a non-zero exit
code being returned from the process.
Closes gh-5922
Previously, if the folder which contained the jar was a symlink the
launch script would use the symlinked folder's name when determining
the default identity.
This commit updates the launch script so that symlinks are resolved
and the canonical name of the folder which contains the jar is used
when determining the script's default identity. The behaviour when
APP_NAME has been set is unchanged.
Closes gh-5679
Closes gh-5733
This commit improves the `InfoContributor` infrastructure as follows:
* `InfoEndpoint` no longer breaks its public API and returns a Map as
before
* `Info` is now immutable
* All properties of the build are now displayed. Since we control the
generation of that file, there is no longer a mode to restrict what's
shown
* Build info is now generated in `META-INF/build-info.properties` by
default
Closes gh-5734
Previously, if a void method with a single argument was named "set", the
annotation processor wrongly considered it was a setter candidate. This
commit updates the condition to ignore it.
Closes gh-5826
Previously, the Gradle plugin would include all of the dependencies
from both the compile and runtime configurations in the repackaged
jar. In the unlikely event that the compile and runtime configurations
contained different versions of the same dependency, this would lead
to both versions of the dependency being packaged in the jar file.
The runtime configuration extends the compile configuration so, in
normal circumstances, it will contain a superset of the compile
configuration's dependencies. In the situation described above where
the two configurations contain different versions of the same
dependency the runtime configuration will only contain whichever
version of the two dependencies has "won". By default, this will
be the dependency with the higher version.
This commit updates the Gradle plugin to only include the runtime
configuration's resolved dependencies during repackaging. As explained
above, the runtime configuration extends the compile configuration so
any compile dependencies will still be included, with the added
benefit that duplicate versions of the same dependency will have been
resolved to a single, preferred version.
Closes gh-5749
Previously, a JVM that was forked by spring-boot:run could be orphaned
when the parent process (the Maven build) was terminated in an IDE. Note
that this doesn’t happen when spring-boot:run is invoked from a shell.
This commits add a shutdown hook that registered when RunMojo forks the
JVM. The shutdown hook attempts to kill the forked JVM’s RunProcess
rather than relying on the death of the parent process being sufficient
to also kill the child.
Closes gh-5815
Update CentralDirectoryParser to reduce the number of objects created
when parsing the central directory. A single CentralDirectoryFileHeader
object is now reused as entries are parsed.
Fixes gh-5260
Previously, the CLI would always use the class loader to try to locate
a source file. This was contrary to the SourceOptions javadoc which
states that the class loader is “used to try and load files that are
not found in the local filesystem”. This provide to be problematic on
Windows when a fully-qualified source file was supplied. The driver
letter colon slash (e.g. c:/) at the start of the path is considered
invalid for a class path resource by URLClassPath.Loader resulting in an
IllegalArgumentException being thrown.
A workaround for this URLClassPath behaviour was added in a71c9b5d. It
was removed as part of reworking LaunchedURLClassLoader to use a
conventional delegation model in 87fe0b2a. It was then reinstated in
cc140b2c. This work around is undesirable as it causes
LaunchedURLClassLoader’s behaviour to diverge from URLClassLoader’s
behaviour (this is contrary to the comments in the test added in
a71c9b5d which incorrectly tests the two class loader with different
class paths. If the two class loaders are created with the same class
path then their behaviour is the same).
This commit updates SourceOptions to make its behaviour match its
javadoc so that a search of the class path is only performed if the
filename doesn’t exist on the filesystem. Furthermore, when running on
Windows, if the filename is an absolute path no further searching is
performed as the path cannot reliably be used to search the class path
due to the behaviour of URLClassPath.Loader when given a path that
is an absolute file path on Windows. This ensures that the user is
presented with an error message indicating that the file could not be
found.
Closes gh-5650
Andy Wilkinson
Johnny Lim
Stephane Nicoll
Spring Buildmaster
Phillip Webb
hengyunabc
Dave Syer
Tommy Ludwig
Christian Flamm
Jakub Narloch