Test failing in kyro-serializer

[wicketstuff]

Test failing in kyro-serializer

[martin-g]

I cannot reproduce this problem locally.
I saw that the error message is wrong (copy/paste error), it should be: Assert.assertEquals("The produced data length is not correct!", 437, data.length);
But for some reason the exception you see doesn't even say what is expected.
Since when you experience this problem ? Or when was the last time you built this project locally ? Because the last logical changes have been made at 18 July.

[akiraly]

The bug only shows if you use RC7 instead of SNAPSHOT (notice the 5 character difference :)).

[martin-g]

hehe. who uses stable versions these days ... :-)
good catch!

[brunoborges]

What about Hudson? Is it working? I've tried to run some builds, but all of them failed...

[mocleiri]

This is still failing as of the 1.5.3 release. Actually it has never worked for me I have just been skipping the tests for the kyro-serializer.

Is there anyway to fix this so its not tied to the ever changing project version?

[mocleiri]

what about a maven profile that activates when -DperformRelease=true and will exclude this test from the surefire plugin?.

[martin-g]

No need to exclude it. Just rework it to not use the version as Label model but something stable.
I thought Attila's fix worked but since it still fails for you then let's rework it.

[martin-g]

Must be better with 9aaa4d3

[akiraly]

Maybe we shouldn't check the exact size (or only that it is greater than some minimum, like 100 bytes) because every new field added to the class hierarchy can break it which makes this test very fragile.

[martin-g]

Agree!

[martin-g]

Fixed with e31c61d.