On 12/10/2019 00:03, Phil Steitz wrote: > How about adding the DBCP unit tests to the source tree? I suspect some > would have failed due to this change. If others think this is a good > idea, I could take a stab at genericising them and creating a PR to add > them.
+1 I'd suggest several commits. Something like: - Copy latest 1.x from Commons - Fix package naming issues - Fix any running issues - Fix warnings (inc. generics) That way it should be easy to trace what changed from Commons and why. The exact detail of how the changes are split between commits is not important. It is the easy traceability that matters. Mark > > On 10/11/19 3:31 PM, bugzi...@apache.org wrote: >> https://bz.apache.org/bugzilla/show_bug.cgi?id=63833 >> >> Phil Steitz <p...@steitz.com> changed: >> >> What |Removed |Added >> ---------------------------------------------------------------------------- >> >> OS| |All >> >> --- Comment #2 from Phil Steitz <p...@steitz.com> --- >> This is a regression from the generics conversion in >> PoolableConnectionFactory. >> >> The original DBCP 1.x code effectively null-checked the object to be >> destroyed: >> public void destroyObject(Object obj) throws Exception { >> if(obj instanceof PoolableConnection) { >> ((PoolableConnection)obj).reallyClose(); >> } >> >> Removing the instanceOf check makes NPE possible: >> public void destroyObject(PoolableConnection obj) throws Exception { >> obj.reallyClose(); >> } >> >> Solution is to add an explicit null check in destroyObject. >> >> Similar changes should be made to activate, passivate, validate methods. >> > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org > For additional commands, e-mail: dev-h...@tomcat.apache.org > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
