On 17/04/2023 19:28, Bruno Kinoshita wrote:
BTW, for this test I deleted my ~/.m2/repository, deleted the `jena` dir
from my workspace. git cloned it again, checked out the 4.8.0 tag, and
(without ever opening Eclipse or Webstorm - used for jena-fuseki-ui) I
executed `mvn clean install > log.txt 2>&1`.
Thanks - the same class of errors - class initialization failing and
that marks the class "not present".
I've recreated it on Windows (only).
Reverting the offending git commit makes it go away.
It looks like it is a class initialization (involving Commons
lang3.SystemUtils, jena.base.Sys and TDB1 tests).
Not completely reassuring that I haven't found the exact fault path but
with the change reverted, a failing setup reliably gets pass the
offending point in the build.
I'm running all the GH actions to check. If that's OK I'll send in a PR.
(There are other ways to fail on GH - there something about the way that
tests end when they have forked a Fuseki server that continues to hold
network-level resources - this isn't new - it has only been on GH actions)
Andy
PS GH Actions looking OK (I'm running them on my account because it
pollutes the git repo history trying things out - MacOS action is
far-and-away the most reliable!)
On Mon, 17 Apr 2023 at 20:26, Bruno Kinoshita <brunodepau...@gmail.com>
wrote:
Same error. Saved the log here (mvn -v output as well):
https://gist.github.com/kinow/ce0435d4ffd1e4a2fcfede53735cd03e
On Mon, 17 Apr 2023 at 17:42, Bruno Kinoshita <brunodepau...@gmail.com>
wrote:
Hi Andy,
I am on Ubuntu 22.04.1 LTS. I will delete my git repo, clone again,
delete my Maven cache, and try again (in a few hours, after Maven has
downloaded half of the Internet).
Cheers
Bruno
On Mon, 17 Apr 2023 at 17:30, Andy Seaborne <a...@apache.org> wrote:
Bruno - what OS are you using?
I triggered all the jobs we have (Jenkins and github) and windows jobs
now show something like what you are seeing.
But also other weird stuff:
jena-core/test;
[INFO] Skip filter: Not( Wildcard( Sensitive, *.test.* ) )
[INFO] Could not create Interface report class
java.lang.IllegalArgumentException: No classes found in
[org.apache.jena.assembler, org.apache.jena.datatypes,
org.apache.jena.enhanced, org.apache.jena.graph, org.apache.jena.mem,
org.apache.jena.ontology, org.apache.jena.rdf, org.apache.jena.rdfxml,
org.apache.jena.reasoner, org.apache.jena.shared, org.apache.jena.util,
org.apache.jena.vocabulary]
that is all within jena-core!
then jena-tdb1:
[INFO] Running org.apache.jena.tdb.assembler.TS_TDBAssembler
[ERROR] Tests run: 7, Failures: 0, Errors: 7, Skipped: 0, Time elapsed:
2.454 s <<< FAILURE! - in org.apache.jena.tdb.assembler.TS_TDBAssembler
[ERROR]
org.apache.jena.tdb.assembler.TestTDBAssembler.createDatasetDirect Time
elapsed: 1.755 s <<< ERROR!
java.lang.ExceptionInInitializerError
. . .
Caused by: java.lang.NullPointerException
at org.apache.jena.rdf.model.impl.ModelCom.add(ModelCom.java:1141)
at
org.apache.jena.assembler.assemblers.AssemblerGroup$ExpandingAssemblerGroup.implementWith(AssemblerGroup.java:106)
https://ci-builds.apache.org/job/Jena/job/Jena_Development_Windows/209/consoleFull
On github - the windows job got past TDB1 then hit a networking/timeout
issue that has been GH specific.
Now it shows the "Could not create Interface report class" then
jena-tdb1: jena-core issue and then:
[INFO] Running org.apache.jena.tdb.assembler.TS_TDBAssembler
[ERROR] Tests run: 7, Failures: 0, Errors: 7, Skipped: 0, Time elapsed:
2.454 s <<< FAILURE! - in org.apache.jena.tdb.assembler.TS_TDBAssembler
[ERROR]
org.apache.jena.tdb.assembler.TestTDBAssembler.createDatasetDirect Time
elapsed: 1.755 s <<< ERROR!
java.lang.ExceptionInInitializerError
...
Caused by: java.lang.NullPointerException
at org.apache.jena.rdf.model.impl.ModelCom.add(ModelCom.java:1141)
at
org.apache.jena.assembler.assemblers.AssemblerGroup$ExpandingAssemblerGroup.implementWith(AssemblerGroup.java:106)
https://github.com/apache/jena/actions/runs/4718624345/jobs/8368433891
Currently, it looks to me to be (1) test related - and some tests do
dive straight into Jena and can bypass initialization (2) something has
changed the hash order
There has been one TDB1 change recently ... but why it affects the build
in a non-deterministic way is difficult to explain.
I'll try some changes and see if the GH action for Windows can be made
to behave differently.
Andy
On 16/04/2023 20:49, Andy Seaborne wrote:
On 16/04/2023 20:09, Bruno Kinoshita wrote:
I wonder if I have to check out from scratch again, or maybe I need to
update Maven or JDK, or use a different command?
I'm trying to build it with Java 17 (OpenJDK) with `mvn clean test
install
-Pdev`, `mvn clean install -Pdev`, and `mvn clean install`. It always
fails
on TDB1, failing to run the tests.
[ERROR] TestTransactionUnionGraph.before:43 NoClassDefFound Could
not
initialize class org.apache.jena.sparql.sse.SSE
@Before
public void before()
{
ds = TDBFactory.createDataset() ;
ds.asDatasetGraph().add(SSE.parseQuad("(<g> <s> <p> 1)")) ;
}
NoClassDefFound means it compiled but then wasn't found at runtime. So
it (SSE) was there ... then it wasn't!
Failing to initialize a class can look like class not found. SSE will
have been used in earlier modules. Strange.
[ERROR] TestTransactionUnionGraph.before:43 NoClassDefFound Could not
initialize class org.apache.jena.sparql.sse.SSE
[ERROR] TestTransactionUnionGraph.before:43 NoClassDefFound Could
not
initialize class org.apache.jena.sparql.sse.SSE
[ERROR] TestTransactionUnionGraph.before:43 NoClassDefFound Could
not
initialize class org.apache.jena.sparql.sse.SSE
[ERROR] TestTransactionUnionGraph.before:43 NoClassDefFound Could
not
initialize class org.apache.jena.sparql.sse.SSE
[ERROR] TestTransactionUnionGraph.before:43 NoClassDefFound Could
not
initialize class org.apache.jena.sparql.sse.SSE
[INFO]
[ERROR] Tests run: 906, Failures: 0, Errors: 484, Skipped: 5
Any idea what's going on?
I've just downloaded the source zip on a machine which wasn't the
release machine. (Linux again)
"mvn clean install -Pdev" worked.
I sometimes get similar-looking problems when Eclipse is running while
running maven outside the IDE.
Eclipse sees things changing and decides to rebuild the world. Eclipse
does a clean ... and deletes maven's earlier work. That might explain
why it was there and then it wasn't. It could also break system
initialization.
Andy
Thanks!
On Sun, 16 Apr 2023 at 19:16, Andy Seaborne <a...@apache.org> wrote:
+1
On 16/04/2023 17:30, Andy Seaborne wrote:
Hi,
Here is a vote on the release of Apache Jena 4.8.0.
This is the first release candidate.
The deadline is
Wednesday, 19th April 2023 at 20:00 UTC
Please vote to approve this release:
[ ] +1 Approve the release
[ ] 0 Don't care
[ ] -1 Don't release, because ...
