I tried to set up Java to test this under a macOS environment but
encountered multiple problems, possibly related to Java 8/12 and gson.
My interest in debugging a non-Linux configuration has reached its
limit; if this situation improves I can look at it again. From my
perspective this should
Looks like the Etag is persisted in the USerDefinedFileAttirbuteView, which
is not supported in OSX. The tests that check that were introduced in
master but not in 2.1.x, that's why we did not see that before when
releasing bugfix branches.
I've added a fix to ignore those checks in OSX,
The failures are weird, it seems that the ETag is not a multipart and
instead a full part. Maybe some issue storing xattr? I think it is OK
to release without this since filesystem is usually used for testing.
On Wed, Oct 09, 2019 at 02:58:10PM +0200, Ignasi Barrera wrote:
> I only have easy
I only have easy access to OSX now, so I was running the release there.
Anyway, if there have been no relevant code changes for quite some time,
and we have not had any issues regarding the filesystem provider in OSX
(which I'd say is not being widely used), I think it is reasonable to not
Unfortunately I do not have easy access to a Mac for a few days. This
does work on Linux and it does not appear that there are any filesystem
or blobstore changes in git log since 2.1.2. Perhaps JVM versions
influence this?
On Wed, Oct 09, 2019 at 12:30:27PM +0200, Ignasi Barrera wrote:
> Hi!
>
Und jetzt noch fixen.
Am 9. Oktober 2019 13:40:01 schrieb "Meyer, Carsten" :
Reproducible on my Mac.
Best regards,
Carsten
Tests run: 109, Failures: 3, Errors: 0, Skipped: 15, Time elapsed: 12.757
sec <<< FAILURE! - in TestSuite
Reproducible on my Mac.
Best regards,
Carsten
Tests run: 109, Failures: 3, Errors: 0, Skipped: 15, Time elapsed: 12.757 sec
<<< FAILURE! - in TestSuite
testMultipartUploadMultipleParts(org.jclouds.filesystem.integration.FilesystemBlobIntegrationTest)
Time elapsed: 0.02 sec <<< FAILURE!