afs commented on PR #3063:
URL: https://github.com/apache/jena/pull/3063#issuecomment-2797876988
Hi @OyvindLGjesdal
We're seeing the same sort of thing.
I think you may be on to something about `conn.update("CLEAR ALL")`.
I can't explain why its always the same test that fails but one server per
test + protected "CLEAR ALL" is working (caveat - its now a quieter time of
day).
I've got some changes that can go to a PR for `main` when cleaned up.
The most annoying failure is `TestFusekiShaclValidation.shacl_no_data_graph`
because it is in `jena-fuseki-main`, whereas the other typical failures are in
`jena-integration-tests` which nothing depends. Running "fail at end" would at
least continue most code checking.
* reinforced the exception handling path to always consume a potential
message and always close the connection,
* Made tests run one server per test (code-change configurable to one per
suite) - this reduces the failure rate to 1 in 10
* Protected `conn.update("CLEAR ALL")` - the error rate on java test
failures is zero
* Move most of the typically failing tests into their own classes for
convenience.
These rates are job failures, runs of both Java17 and Java21, so the
per-build twice as good (it is very rare to have both fail).
The UI issue looks to be an unrelated random failure.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
