On 10/04/2014 02:34, Christopher Schultz wrote: > Mark, > > On 4/8/14, 10:55 AM, Mark Thomas wrote: >> On 08/04/2014 06:32, Emmanuel Bourg wrote: >>> Le 08/04/2014 14:03, Mark Thomas a écrit : >>> >>>> Can you provide the details of the failure. That might help >>>> ID a solution. >>> >>> Thank you, here is the build log on Debian: >>> >>> http://87.98.165.193/debian/openjdk8-rebuild/logs-failed-jdk8/tomcat7_7.0.52-1_unstable_jdk8.log >>> >>> >>> [javac] >>> /«PKGBUILDDIR»/java/org/apache/catalina/tribes/tipis/AbstractReplicatedMap.java:1107: >>> >>> error: keySet() in AbstractReplicatedMap cannot override keySet() in >>> ConcurrentHashMap [javac] public Set<Object> keySet() >>> { [javac] ^ [javac] return type >>> Set<Object> is not compatible with KeySetView [javac] where >>> K,V are type-variables: [javac] K extends Object declared >>> in class ConcurrentHashMap [javac] V extends Object >>> declared in class ConcurrentHashMap >>> >>> >>> This issue is caused by the addition of a keySet() method to >>> ConcurrentHashMap in Java 8 with a return type incompatible >>> with the keySet() method in AbstractReplicatedMap. >>> >>> In r1222329 AbstractReplicatedMap was changed to no longer >>> inherit ConcurrentHashMap and implement directly the Map >>> interface. This had the side effect of fixing this Java 8 issue >>> for Tomcat 8. >> >> That looks like something we are going to need to fix but any fix >> is going to have to change the API. On that basis, back-porting >> r1222329 is probably the best option. >> >> Any objections to back-porting this fix to 7.0.x? > > +1 to back-porting
I spoke with Filip at ApacheCon. This is meant to be an internal API so the refactoring should be safe. I've done the back-port and fixed a handful of IDE warnings while I was at it. Mark --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
