Re: lucene2

[Emmanuel Bourg <ebourg@apache.org>]

Le 01/08/2014 16:59, Stephen Nelson a écrit :

> I wonder if the newer JDK is triggering this instability
> in the tests? Otherwise with this amount of instability, bugs would
> have been raised against upstream when this code was current.

If the failure occurs with Java 8 it might be caused by the new
implementation of the HashMap class. The order of the entries is not
guaranteed in an HashMap but it was rather stable and tests were written
assuming the order wouldn't change. This is no longer true with Java 8,
and the tests have to be fixed.

Emmanuel Bourg