OyvindLGjesdal commented on PR #3063:
URL: https://github.com/apache/jena/pull/3063#issuecomment-2727633929
Hi @afs
I think you are right that the tests I added are not related to the failures.
It looks to me that (one of) the errors is that
HttpLib.exception(httpResponse, httpStatusCode) can sometimes cause failures.
My understanding that I’m not confident in:
The `TestFusekiShaclValidation.shacl_no_data_graph` failure seems to go
away by removing usage of the method `HttpException exception(httpResponse,
httpStatusCode)` in ` `void handleHttpStatusCode(HttpResponse<InputStream>
response)` in the HttpLib class, and replacing it with the exception handling
and `void finish(httpResponse<InputStream>)` that is also used for the case of
300-399 return codes. This removes use of the httpResponse body in the returned
exception message, so this is for experimenting, and no fix. I also see other
errors sometimes that may not have been there before. The exception method is
only used in these cases for 400-499 and 500-599. Could consume
Most notable difference to me is that
`finish(hr)->finish->(in)->consume(final in)` does not close the `in` and the
`exception method does (using finally { IO.close(in); }`).
I *think* this is the steps in `shacl_no_data_graph` when the test fails:
* `expect404(…)` passes.
* `conn.update("CLEAR ALL»);` then tries to run, but the connection is
already closed (sometimes).
## After the change
There are also some other new(?) CI errors that then show up, like in
TestFusekiStdReadOnlySetup.stdSetup_endpoint_7, which returns:
```
TestFusekiStdReadOnlySetup.stdSetup_endpoint_7:112 PUT
http://localhost:41455/ds/data :: ==> expected: <405> but was: <-1>
```
## Other things
* It looks like the function copied from commons-io for consume has new
code, I tried it, but it made no difference. Have reverted.
* There was a similar error message in selenium
https://github.com/SeleniumHQ/selenium/issues/11798 caused by a bug in the
HttpClient which was turned into a JDK bug and fixed in Java22
(https://bugs.java.com/bugdatabase/view_bug?bug_id=8304701). Using the
workaround of setting Redirect.NONE in the default http client had no effect
afaik. Could still test if current main runs without issues on 22 in
github-actions.
--
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]
