[11u]: Backport of RFR 8211122: Reduce the number of internal classes made accessible to jdk.unsupported (and JDK-8205537 and JDK-8211121)

[Andrew Haley]

On 6/24/19 2:51 PM, Langer, Christoph wrote:

> please help me reviewing the backport of “JDK-8211122: Reduce the
> number of internal classes made accessible to jdk.unsupported”. The
> main reason for backporting this item is that it’ll ease further
> backports which base on that changeset. The patch is quite extensive
> and hence doesn’t fully apply. I had to resolve some rejects and
> also modify a few other places. Lots of the rejects were about
> copyright years (especially in the hotspot tests) which could be
> dropped. Find below the link for a full webrev and an incremental
> one which only contains my manual changes. I’m currently running it
> through SAP’s test system to check for regressions.
> 
>  
> 
> Though this change is large and thorough review is tedious, I think
> the main issue is that we need 2 additional fixes to pull in before
> taking 8211122:
> 
> JDK-8205537: Drop of sun.applet package
> 
> JDK-8211121: Remove sun.reflect.ReflectionFactory::newInstanceForSerialization
> 
> As the names already suggest, both changes remove
> methods/classes. I’m not 100% sure whether these removals should be
> done in a JDK11 update releases. I rather think they could because
> the one is about Java applets which aren’t supported with (Open-)JDK
> 11 anyway and the other removed method claims to only have been
> added in 9.0.0.4 to be used by the meanwhile removed CORBA modules
> (removed with JDK11). But I’d really like to have some opinions
> about these.
> 

I'm nervous. Although I expect this won't cause any JCK compatibility
issues, it is a behavioural change. I wonder if we should consider
referring this one to the CSR team.

-- 
Andrew Haley  (he/him)
Java Platform Lead Engineer
Red Hat UK Ltd. <https://www.redhat.com>
https://keybase.io/andrewhaley
EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671