Use special Maven settings with access token to GitHub Packages
The access token is for the 'kriegaex' account. Can be adjusted or
extended in order to support other Package registries, too. for now, I
just want to see it this solves the authentication error problems during
GitHub CI builds.
The new file .mvn/settings-read-github-packages.xml contains additional
information and links to online sources, explaining why this is
necessary.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Before, the Maven Clean configuration overrode the one from the parent
POM. Now it leaves it intact, adding a separate module-specific
execution to delete the downloads and libraries.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Make sure to clean up temp-dirs in 'weaver' module
Maven Clean now deletes '' directories if it finds any. Furthermore,
AsynchronousFileCacheBackingTestSupport now not just deletes directory
contents but also removes the empty corresponding directories
afterwards.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
On GitHub CI, there is a very strange error while downloading the POM,
which does not occur locally. Maybe this is due to the usage of
inline XML tags inside a CDATA section in the 'description' tag text.
The default mode removes the description.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
There is a warning because 'asm-renamed' uses
<version>${asm.version}</version> in its artifact descriptor instead of
a fixed version. but as long as Maven still permits it, let us use it
this way. Flatten Maven plugin replaces it by a resolved number anyway
for the dependency-reduced POM.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
There were some problems in file handling: One file in was not deleted
in case an exception was thrown during the test. Another case was a
JarFile which was not closed before deletion, which might work on Linux,
but not on Windows where the open file is still locked after usage.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Module 'asm-renamed' now deploys to GitHub Packages
This means that instead of a system-scoped dependency we now have a
regular one.
The 'libx' module also downloads binary and source JARs redundantly into
the libraries directory in order to be found there by other scripts and
tests.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Replace JDT Core system dependency by deployed one
Get rid of system paths. Instead, rely on JDT Core Shadows to deploy
both binary and source JARs to GitHub Packages. The former module
directory was deleted completely. Instead, the JARs are redundantly
copied into 'libs/jdtcore-aj' in order to be found there by tests and
other Ant scripts.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Build libs without additional profiles, add Enforcer rule
Because 'libx' no longer is a submodule of the AspectJ parent POM, it
will not be built automatically each time an AspectJ build runs.
Therefore, is is no longer necessary to shield the zip/unzip steps from
repetitive execution by profiles with auto-activation based on the
(non-)existence of files. An AspectJ developer knows when to build the
module, she does it manually anyway.
A new Enforcer rule makes sure to warn the developer if some files it
expects to exist in the libs folder are not actually present.
Now we also have a Maven Clean rule which wipes away all downloaded and
(un-)zipped files.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Enable CI build to be run manually and add download libs step
Via 'workflow_dispatch' users with the necessary access rights can now
run the GitHub Actions workflow from the web UI.
Still in testing stage in redundant module 'libx', prepare for the
future situation that currently committed binaries in 'lib' shall be
replaced by downloaded ones.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Remove ASM 2.0 dependency from AtAjLTWTests::testLTWUnweavable
The test class UnweavableTest used ASM 2.0 API. I upgraded in two ways:
1. Now the ASM 9.1 API is used. Probably works with much older
versions too (just not as old as 2.0), as long as the method and
constructor signatures are the same).
2. The class now uses the AspectJ version of ASM (i.e. package names
aj.org.objectweb.asm.*) and therefore can just use ASM as it is on
the classpath for module 'tests' already. There is no more need to
manually add '<pathelement path="${aj.root}/lib/asm/asm-2.0.jar"/>'
to the Ant build script for that test.
Consequently, asm-2.0.jar can be eliminated from Git SCM completely,
because it was only used in this one test.
BTW, I also removed some deprecated API and other types of warnings in
UnweavableTest.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Replace system-scoped dependency on commons by granular dependencies
There are only two direct dependencies used in AspectJ code:
- Commons Digester (module 'testing')
- Commons Logging (module 'org.aspectj.matcher')
I declared those two and experimentally removed all the other
system-scoped dependencies, as it should be. Let's see if the build
works with transitive dependencies.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Recreate lib/commons from Apache Commons downloads
Project archeology, binary and source code comparisons of contents in
lib/commons/commons.jar and lib/commons/commons-src.zip yielded the
following results:
- All binaries are available on Maven Central in 4 different legacy
Apache Commons dependencies:
* commons-beanutils:commons-beanutils:1.4
* commons-collections:commons-collections:2.0
* commons-digester:commons-digester:1.3
* commons-logging:commons-logging:1.0.1
- Those Maven Central binaries are not accompanied by source JARs,
i.e. in order to recreate lib/commons/commons-src.zip we have to
download source archives from the corresponding Git tags. All
projects are available on GitHub, so it is possible to download them
using Maven Download Plugin.
- Both the compound binaries and compound sources archives currently
checked in in AspectJ can be recreated using TrueZIP Maven Plugin.
This is rather tedious and involves additional Maven profiles in
order not to generate the compound archives during every build, but
fully implemented now.
Unfortunately, all of the above does not make the system-scoped
dependency on commons.jar obsolete. In order to achieve that, we either
have to publish the compound files on Maven Central or GitHub Packages,
or we find out which AspectJ modules use classes from which of the 4
individual Apache Commons packages and replace the compound system
dependency by the relevant single dependencies. Probably I am going to
try that in a next step.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Recreate lib/ant from Apache source/binary downloads
- Download Ant 1.6.3 binaries and sources ZIPs from Apache releases
download server
- Verify expected SHA-1 checksums
- Unpack binary distribution
- Repack main sources into source package as it is checked in now
- Redundantly add JUnit JAR in order to 100% replicate existing
directory layout
- Move downloads from 'validate' phase to 'generate-resources'
- Unpack/repack phase is 'process-resources'
- Make sure that download, unpack, repack only occur if necessary
instead of overwriting existing artifacts during each build
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Unzip dependencies in phase 'prepare-package' before building assemblies
Before it was phase 'validate', which was way too early and somewhat
annoying and time-consuming when during development we just call
validate in order to check if the POMs are valid, as the name implies.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Add developer info + sample config about how to work with a RAM disk
There are two files:
- docs/developer/ram-disk/maven.config
- docs/developer/ram-disk/settings-ramdisk.xml
The latter contains info about how to set up a development environment
inside a RAM disk. Both files are to be copied to the project's '.mvn'
folder in the root directory and adjusted according to the description.
Just in case, .gitignore ignores the files if they exist in '.mvn', so
they are not being staged and committed accidentally.
An additional screenshot shows how to configure the Windows Recycle Bin
to immediately delete files in order too save space on the RAM disk.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Make all tests run on Java 16 via '-add-opens' JVM option
Due to JEP 260 (Encapsulate Most Internal APIs), aspect weaving on
Java 16 now requires '--add-opens java.base/java.lang=ALL-UNNAMED' on
the command line. Otherwise there will be illegal access exceptions for
some internal API calls AspectJ needs, most prominently when trying to
define classes in other packages or modules.
This had to be done on several levels:
- Maven Surefire: running tests in a JVM directly forked by Surefire.
In order to make this backwards compatible, I added two profiles
with JDK-level-dependent auto-activation, one 8-15 and one 16+. In
the latter a property containing the JVM parameter is defined, in
the former it is empty, i.e. the JVM is started without the
parameter. In Java 8 the parameter did not even exist, in Java 9+ we
could use it, but we need to test how users use AspectJ.
- RunSpec: Whenever an XML test is declared to use '<run>', we need to
determine the current JVM version and again dynamically add the
parameter when forking the target JVM.
- AntSpec: Whenever an XML test is declared to use '<ant>', we need to
determine the current JVM version dynamically add two properties
usable from within Ant scripts: 'aj.addOpensKey' and
'aj.addOpensValue'. Unfortunately, Ant needs to use two '<argLine>'
parameters, because the two parts of the option are separated by a
space character.
- Ant scripts: When triggered by an AntSpec, each Ant target using LTW
needs to manually set
<jvmarg value="${aj.addOpensKey}"/>
<jvmarg value="${aj.addOpensValue}"/>
for each '<java>' task. It was quite tedious to find all(?) of them.
TODO: In the AspectJ 1.9.7 release notes we need to document that this
parameter is now needed for LTW.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Before Java 16, JDK proxies were given a virtual package name of
'com.sun.proxy'. Now the packages are numbered 'jdk.proxy[n]', i.e.
'jdk.proxy1', 'jdk.proxy2' etc. This makes the package-name-derived path
name here less predictable. In our simple runtime scenario, we can be
pretty sure than the counter starts at 1 because it is the first and
only proxy we create.
TODO: A better solution would be a recursive filtered search via
Files.walk, ideally added as a recursive search option for
CountingFilenameFilter.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Deactivate test run in 'run-all-unit-tests' module by default
There is a new Maven profile 'repeat-all-unit-tests', and the name
already implies what a comment in the Maven module explains like this:
ATTENTION: This profile is inactive by default, because when active and
running a full reactor build, it makes almost all tests run 2x, doubling
the build time without any added value. This Maven module only exists
for convenience: As a developer, your IDE can detect and run
'RunTheseBeforeYouCommitTests'. That way, you do not have to use Maven
and get the test results reported within the IDE's JUnit user interface.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Add Java 16 Linux matcher variant to HtmlDecorator
The Java 16 Javadoc generator has changed the HTML structure once again
even compared to Java 15. I adjusted the matching in HtmlDecorator and
also fixed CoverageTestCase. Most methods there I just made to work
quickly, but method 'testInnerAspect()' I actually refactored. Some
other methods could (probably should) be restructured in a similar
fashion, but for now I just wanted to understand what the test does and
see how much work it would be to refactor it. But finally, I just want
to get the GitHub CI build running on Java 16.
TODO: I did not check if the decorated HTML actually looks OK and am
unsure if the tests cover that sufficiently, I never reviewed the tests.
It would also be better to do regex matches instead of looking for
variants of fixed strings or maybe even to operate on a DOM. But I am
not in a mood to refactor that tonight.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Add diagnostic output to HtmlDecorator if AJ-Doc generation fails
HtmlDecorator.decorateHTMLFile is where after Java version upgrades
(i.e. also new Javadoc generator version) usually tests fail for the
first time during builds because strings no longer match as expected.
There now is this log message on stdOut: "Something unexpected went
wrong in HtmlDecorator. Here is the full file causing the problem:"
After that, a full HTML page is logged. I hope this helps me identify
the new error on GitHub Linux Java 16, because the same test works on
Windows and I have no idea how to remote-debug a GitHub CI build.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Make BCEL classpath utility recognise Java 16-19, fixing many tests
This is a follow-up commit on @07af5d41:
Inside org.aspectj.apache.bcel.util.ClassPath.getClassPath(), some JVM
version matching occurs which previously did not include Java 16 (I also
added 17-19 to the regex matcher). This fixes test errors like:
java.lang.ClassCastException:
class org.aspectj.weaver.MissingResolvedTypeWithKnownSignature
cannot be cast to class
org.aspectj.weaver.ReferenceType
(org.aspectj.weaver.MissingResolvedTypeWithKnownSignature and
org.aspectj.weaver.ReferenceType are in unnamed module
of loader 'app')
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
After Ajc usage text output is filtered into its own category
IMessage.USAGE now - see commit @31b2d60b - some tests in module
'org.aspectj.ajdt.core' were failing. I fixed and also improved them a
bit in @e4a2a5a5, but forgot to commit this one.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Set explicit en_US locale for build timestamps in Build Helper Maven
There were warnings that builds were dependent on the system local
(de_DE in my case). In patterns like "EEEE MMM d, yyyy", parts like day
of week or month names would change otherwise.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
After Ajc usage text output is filtered into its own category
IMessage.USAGE now - see commit @31b2d60b - some tests in module
'org.aspectj.ajdt.core' were failing. I fixed and also improved them a
bit.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Add Java 16 test suite for AspectJ 1.9.7 + test refactorings
- Test all features which were preview in 14+15 and are now final in 16,
compiling them with language level 16.
- For Java 15 we only have sanity tests (and of course the Java <14
tests), compiling Java 16 features to target 15 does not seem to work.
- Test remaining Java 16 preview feature (sealed classes).
- Instead of overriding runTest(String) in several base classes like
XMLBasedAjcTestCaseForJava*Only or XMLBasedAjcTestCaseForJava*OrLater,
we now override setUp() from JUnit's TestCase base class. This will
run before runTest(String) and make the tests fail much faster, if a
user tries to run them on the wrong VM.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Remove module 'run-all-junit-tests' from root POM -> speed up the build
This module must be a relic from a test runner module once existing
during the Ant build era, but transferred and kept alive in the Maven
build. Actually, it almost doubles build time by running virtually all
tests in all modules again when doing 'mvn test' from the project root.
For now I only removed the module from the root POM, leaving behind
comments there, in the module POM and in the now @Deprecated class
RunTheseBeforeYouCommitTests.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
It failed with "RuntimeException: I never heard about what kind of build
it was!!" on my (@kriegaex) Windows machine, mostly because in case of a
failing full build the corresponding status is never set.
TODO: Ensure that 'MyStateListener.informedAboutKindOfBuild' is set for
failed builds, too.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Fix tests not finding project base directory 'org.aspectj'
Several LTW tests using class TestServer failed on my machine because
there was a hard-coded project base directory name 'org.aspectj' in the
class. This class has several other problems, but my quick fix for now -
I did not want to rename my project base directory - was to match on a
regex '(?i)(org[.])?aspectj' now. That also works for my root directory
'AspectJ'.
I also moved the code determining the project dir into a protected
(hence testable) method and added a sanity test case checking if the
directory can be determined. If not, the test will fail with a rather
lengthy warning to developers about the need to have a matching project
root folder.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Fix tests expecting usage texts as failure outputs
This is a follow-up on commit @31b2d60b. Some tests were actually
expecting usage texts as failure outputs. Because that was fixed, the
tests no longer see those failures, hence they should no longer expect
them.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Fix test relying on JVM warning being stripped from output
The line in which warnings like "Archived non-system classes are
disabled because the java.system.class.loader property is specified"
appears can start with e.g."OpenJDK 64-Bit Server VM" or "Java
HotSpot(TM) 64-Bit Server VM". Therefore, an exact match on the former
worked on Linux, but not on Windows, or maybe the difference is
generally between Oracle and OpenJDK. anyway, I use Oracle on Windows
and my build failed. Now it is fixed because I made the match more
generic using a regex.
I also removed a now obsolete check for the occurrence of the stripped
line in test "JDK14 LTW with XML".
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Improve usage text, error and warning output in batch compiler
- Usage texts are now printed to stdOut, no longer stdErr.
- 'java ...Main -?' no longer prints usage text twice (once to stdOut
and then again to stdErr).
- AjdtCommand.inferKind: Usage texts are no longer mis-identified as
warnings or errors just because they contain substrings "warning" or
"error". Matching is now more precise, looking for "[warning]" and
"[error]". But in that case the method would not be called anyway
because errors and warnings are identified in other ways already. As a
fall-back, the categories IMessage.ERROR and IMessage.WARNING still
exist in the method.
- In case of compile errors, no usage message is printed anymore,
because previously the user had to scroll up a lot in order to see the
actual messages. This is also in line with ECJ. The same is true for
warnings, but it was like this in Ajc already.
- AjdtCommand.inferKind: There is a new category IMessage.USAGE
especially for AspectJ usage texts, which will be identified by string
matching and then correctly handled (i.e. printed to stdOut, not
stdErr).
- Usage text printing is no longer done in AspectJ but in the AspectJ
"shadows" fork of JDT. This helps to get rid of some now obsolete code
here.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
- Use spaces instead of tabs for indentation.
- Document parameters bound in Eclipse JDT, e.g. {0} is not the compiler
name there but the system's path separator ';' or ':'. So if we want
to display the compiler name, we need {1}.
- For both normal usage and '-X' usage, compiler name + version are
printed now.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Fix invalid manifests leading to "Could not find or load main class"
WARNING: Please avoid line breaks in manifest values! They are passed on
like this: Assembly Plugin -> Plexus Archiver -> JRE java.util.jar
.Manifest.write(OutputStream).
The JRE Manifest class inserts hard line breaks always after 72
characters, no matter if those 72 characters contain line feeds, tabs or
spaces. Hence, it can happen that unwanted blank lines end up in the
middle of a manifest section, making the manifest invalid. Calls like
e.g. 'java -cp aspectjtools.jar org.aspectj.tools.ajc.Main' can then
fail with the absolutely unexpected error 'Could not find or load main
class org.aspectj.tools.ajc.Main'.
In IntelliJ IDEA you can deactivate wrapping text inside XML tags like
this: "File | Settings | Editor | Code Style | XML | Wrap text -> off"
The problem occurs in Maven Assembly in versions higher than 2.2. More
exactly, it occurs in Plexus Archiver after in more recent versions it
switched to using the JRE Manifest class.
TODO 1: Either add a test step in phase 'verify' doing something like
this:
new Manifest(new FileInputStream("MANIFEST.MF"));
This would lead to "IOException: invalid header field (line xy)" in case
of an invalid manifest file.
TODO 2: Or file a JRE bug at Oracle or OpenJDK, wherever appropriate.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Strip down compiler messages to AspectJ-specific ones
Delete all properties from messages_aspectj.properties which were just
copied from Eclipse and basically the same. This not only gets rid of
duplicates but also eliminates differences found between upstream and
AspectJ strings which were just cause by errors or oversights during
manual upgrade.
TODO:
- Find a way to print the '-X' options as info instead of yielding
'abort', making it seem as if compilation failed and print the usage
message to stdErr instead of stdOut.
- Eclipse also has misc.usage.warn, not just misc.usage, i.e. usage
info specifically for warning options. Make sure that AspectJ uses
it consistently.
- Find a way to merge AspectJ-specific options into the standard
Eclipse usage text instead of completely replacing it, further
reducing the need to merge and copy upstream content. Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Begin migration to 'aspectj' locale for compiler messages
Renamed messages.properties to messages_aspectj.properties and moved to
a path identical with the Eclipse original resource file. Therefore, it
is now just treated as a localised version of it.
The new jdtcore-for-aspectj.jar already contains logic to use the new
locale. Hence, BuildArgParser no longer needs the static block to
initialise its own resource bundle, Eclipse JDT will automatically pick
it up.
The version string was also updated to the new Eclipse merge commit
hash + date + "Java16".
TODO: Strip down properties file in order to only contain delta to
Eclipse.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Do not run tests using Java 14 preview features on this Java 15 compiler
These tests need a Java 14 level AspectJ compiler, because they use
version-specific preview features. This compiler has been upgraded
to a Java 15 compliant JDT Core already, i.e. it does not support
preview features of a previous version anymore.
An error message similar to the above explanation will appear when
trying to run any XMLBasedAjcTestCaseForJava14Only subclass, such as
Ajc196PreviewFeaturesTests (currently the only one).
When running AllTestsAspectJ196, Ajc196PreviewFeaturesTests will not be
added to the test suite anymore.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Keep only ASM 2.0 binary because it is still used in UnweavableTest
which uses an old ASM API, e.g. with a ClassWriter constructor which no
longer exists.
Also add JarJar 1.3 library because it is needed by an Ant task in
lib/asm/build.xml.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Simplify execution of PureJavaTests and KnownLimitationsTests
Getting rid of XML includes and two superfluous files, merging them with
their respective including counterparts. As far as I can see, the two
test suites are not part of the automatic build process, but can be
started manually as easily as any other test (suite) now.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Fix outdated ajcTestSuite.dtd and lots of XML problems
Now there should be no more inspection warnings when working with XML
test definitions. Only the strangely looking XML files used by
PureJavaTests and KnownLimitationsTests are left to be analysed and
fixed.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Make sure Java 14-only tests using preview are not executed on JVM 15+
Some Ajc196 tests are using preview features (see .../ajc196.xml), i.e.
they will fail on Java 15+ because code compiled with '--enable-preview'
can only run on the same JVM version, not on a more recent one. Hence,
the preview-using tests are now being excluded in order to make the
build run on Java 15, even though no Java 15 features are present in the
current 1.9.7 snapshot.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Move documentation generation into Maven profile 'create-docs'
The profile is active by default, but can be deactivated via
'-P !create-docs' on the Maven command line. This helps to speed up
'mvn install', maybe in combination with skipping tests, if we just want
to create a new set of AspectJ binaries.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Improve text matching in OutputSpec, fixing some failing Windows tests
Some Java 14 text block tests failed on Windows because a
StringTokenizer was used to split by LF, but the Windows line
separator is CR+LF. Because a multi-line string ending with CR+LF is
printed via 'System.out.println' in the test code, another CR+LF is
added to the output, resulting in trailing CR+LF+CR+LF. Hence, between
the two LFs, the tokenizer actually found an additional line consisting
of CR (only on Windows, of course). Despite each line token actually
containing a trailing CR token, that did not matter much because
'String.trim' was used everywhere before comparing values.
Anyway, the improved OutputSpec uses text.trim().split("\\s*\n\\s*"),
which takes care of leading/trailing whitespace both around the whole
output and for each separate line.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Make MoveInstructionsWeaveTestCase.testFancyHello pass when run twice
Reset counter, just in case this test runs multiple times in one JVM.
This can happen e.g. due to "run all tests" in IntelliJ IDEA, which
directly runs this test class and als WeaverModuleTests, both of which
implement junit.framework.TestCase. In that case, during the second run
the counter would start at a higher base count, making the 2nd test run
fail.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Make BCEL classpath utility recognise Java 15, fixing many tests
Inside org.aspectj.apache.bcel.util.ClassPath.getClassPath(), some JVM
version matching occurs which previously did not include Java 15.
Technically, AspectJ 1.9.6 does not support Java 15, but on GitHub
Actions there is a build job running on a JVM 15. This change should at
least make the weaver tests pass, making that test job more meaningful.
This fixes test errors like
java.lang.ClassCastException:
class org.aspectj.weaver.MissingResolvedTypeWithKnownSignature
cannot be cast to class
org.aspectj.weaver.ReferenceType
(org.aspectj.weaver.MissingResolvedTypeWithKnownSignature and
org.aspectj.weaver.ReferenceType are in unnamed module
of loader 'app')
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Fix: WeaveSpec no longer ignores the '-Xlintfile' parameter
Some tests in ajc150.xml and ajc190_from150.xml contain '<weave ...>'
build steps with 'xlintfile="..."' parameters. Those parameters were
passed through to Java and aspect code building steps, but not to the
final weaving step, sometimes leading to spurious "type not exposed to
weaver" warnings which occurred for some local and CI builds, but not
always. Very strange indeed. Anyway, by making method
WeaveSpec.buildWeaveArgs() pass on this parameter, the tests seem to run
reliably now.
TODO: Why does Ajc report that warning if the application JAR is on the
inpath and the aspect JAR is on the aspectpath? Is it because a marker
annotation is defined within the aspect JAR? But actually, that should
not matter, especially not work once and fail at other times. I guess
there is a class loading order problem or similar involved.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Add missing '-options' values -1.5, -1.9 to ajc{150,190_from150}.xml
When trying to find out why ajc150.xml and ajc190_from150.xml look
almost identical, I found out that only in many cases the 190 version
has '-option="1.9"' set where in the 150 version it was '-option="1.9"'.
Unfortunately, in both files source/target versions are not set at all
in many places, which looks unintentional. I tried to search & replace
all relevant '<compile ...>' commands for 190 first, then ported back to
150. Only cases in which clearly compiling to an older version like 1.3
or 1.4 is necessary were left like before.
I am expecting missing or false Java target versions in many other of
the legacy XML test suites. For now, I am just committing these two.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Set maven-assembly-plugin version 3.1.1 explicitly for each POM module
Background: There are CI tests failing because the version is not taken
from the parent POM, even if set in the 'pluginManagement' section.
Instead, it is resolved via the Maven Super POM, e.g. 2.2-beta-5, see:
https://maven.apache.org/ref/3.6.3/maven-model-builder/super-pom.html
That assembly plugin version in turn requires plexus-archiver
1.0-alpha-12. The latter cannot be downloaded from Maven Central,
leading to the problems seen during builds.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Replace path separators ';' by ',' in XML test specs
The goal is for them to be canonicalised to platform standard during
test execution. I am not sure if that will fix any tests, but at least I
hope it will not break any.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Replace directory separator '/' and surrogate path separator ',' by
platform-specific separators File.separatorChar and
File.pathSeparatorChar, respectively. Also make sure that replacement
occurs during write access, not read access. This was handled
differently in both sibling classes.
I am not sure if that helps to fix any Linux CI tests, but it is worth a
try.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Try to fix MultiProjectIncrementalTests.testAspectPath_pr265693
Check if path vs. package name discrepancy makes test fail on Linux. On
Windows it passes. So let's find out if file p/Asp.java vs. pkg.Asp
causes the problems.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
This made ModuleTests.testBuildModuleAndApplyAspectsFromAspectPath fail
because a file delete job for a module JAR failed after a previous
compile job had called FileUtil.isZipFile(File) in which the opened zip
file was never closed. A try with resources fixes that.
Maybe the corresponding test worked on Linux before, I did not try. I
just know that Linux is more forgiving about deleting open files while
on Windows they are being locked, which makes Windows the better system
to search for open file leaks.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
AjBuildManager: use try with resources in 2 places
I was debugging something in ModuleTests, trying to find out if there
are resource leaks. They were not in this class but in FileUtil - see
next commit. However, the little refactoring here does not hurt either.
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
Support Windows 10 and Windows Server 2016/2019 in installer
Those versions were not detected until now, which lead to bogus Windows
batch files forwarding only 9 Ajc parameters to the Java process via
"%1 %2 %3 %4 %5 %6 %7 %8 %9" instead of "%*".
Signed-off-by: Alexander Kriegisch <Alexander@Kriegisch.name>
projects / aspectj.git / log