[11u] OpenJDK 11.0.3 post GA fixes JDK-8210739 and JDK-8203679

[Severin Gehwolf]

Hi Christoph,

On Tue, 2019-04-16 at 08:02 +0000, Langer, Christoph wrote:
> Hi,
> 
> we had already discovered JDK-8210739: Calling JSpinner's setFont
> with null throws NullPointerException [1], [2] as being fixed in
> 11.0.3-oracle b31, which is probably a post 11.0.3 GA build of
> Oracle.
> 
> Another bug has been brought to our attention that had had been fixed
> by Oracle in an 11.0.2 post GA build b31: JDK-8210483: AssertionError
> in DeferredAttr at setOverloadKind caused by JDK-8203679 [3]. The
> issue with that bug is that it was never pushed to the jdk11u
> repository. Oracle has only fixed this in their internal repos.
> Unfortunately, it fell through all the filters because these believed
> it was already part of the open 11.0.2 code base and hence we didn't
> include it in OpenJDK 11.0.3. It has, however, been backported to
> 11.0.4, that is, it was pushed to jdk11u-dev.
> 
> So, both bugs (JDK-8210739 and JDK-8203679) will be missing in the
> imminent jdk-11.0.3+7, respectively jdk-11.0.3-ga pushes.
> 
> I suggest we push both of these fixes (which apply cleanly and are
> pretty local to their area) to jdk11u after Andrew (Hughes) has
> pushed the CPU changes and after that add a jdk-11.0.3+8 tag. Then,
> downstream vendors can chose to create another update based on the
> new tag if they need to deliver binary versions that include these
> fixes. I'll also run these fixes through our SAP test system but I
> don't expect any regressions.
> 
> Any opinions?

Here is one :)

JDK-8210739: Calling JSpinner's setFont with null throws NullPointerException

This bug is present in JDK 9 and better. It's a P3 bug. It's been
around for a long time.

JDK-8210483: AssertionError in DeferredAttr at setOverloadKind caused by JDK-8203679

This bug is present in JDK 11 ea b18 and onwards. It's a P2 bug, but
got filed on September 4, 2018. So it has been around for a while too.

I don't see why either of the two would have to be in any asynchronous
release upstream. Shipping it with the 11.0.4 release in July should be
sufficient for both, IMO. Post the 11.0.3 freeze, regular jdk11u-dev =>
jdk11u merges will happen, EA builds[i] produced. So if somebody really
needs the fix urgently they might go with an EA upstream build. That's
my understanding of the upstream situation.

Downstream vendors, on the other hand, can pick-and-choose however they
like. If they find certain bugs critical enough, they can spin their
own releases.

So to me this is business as usual. There will always be bugs which
will slip releases. With the current Oracle JDK vs OpenJDK situation
there cannot be a 100% match between the two no matter how hard we try.
It'll be close, sure, but not a 100% match.

My $0.02

Thanks,
Severin

[i] http://mail.openjdk.java.net/pipermail/jdk-updates-dev/2019-April/000933.html

> I've flagged both bugs with jdk11u-critical-request. If I get
> maintainer approval by Andrew (Haley), I would do this.
> 
> Thanks
> Christoph
> 
> [1] https://bugs.openjdk.java.net/browse/JDK-8210739
> [2] https://mail.openjdk.java.net/pipermail/jdk-updates-dev/2019-April/000920.html
> [3] https://bugs.openjdk.java.net/browse/JDK-8210483
>