JDK 9 RFR of JDK-8035076: Pattern$BnMS never used due to bug in Pattern$BnM.optimize

[Martin Buchholz]

I don't see an actual test added here:

http://cr.openjdk.java.net/~sherman/8035067/webrev/test/java/util/regex/RegExTest.java.udiff.html


On Wed, Feb 19, 2014 at 2:31 PM, Xueming Shen <xueming.shen at oracle.com>wrote:

> Hi,
>
> Please help codereview the change for JDK-8035076.
>
> Issue:    https://bugs.openjdk.java.net/browse/JDK-8035076
> Webrev: http://cr.openjdk.java.net/~sherman/8035067/webrev
>
> This is regression caused by the change we made back to jdk7 to support
> case
> insensitive match, in which a base class SliceNode for Slice family was
> added and
> we mistakenly subclass the SliceS class to this newly added class, instead
> of
> the original Slice class. The BnM optimization for supplementary support,
> which
> is based on the "instanceof Slice", is therefor disabled. Below is the
> original
> webrev for that changeset.
>
> http://cr.openjdk.java.net/~sherman/6486934_6233084_6504326_6436458
>
> The proposed change is to subclass SliceS from Slice node. The change has
> been
> verified via the test case attached in the issue report. However I decided
> not to
> include this test as a regression test as it is too "implementation detail
> dependent".
>
> Thanks!
> -Sherman
>
>
>