PicoContainer
  1. PicoContainer
  2. PICO-261

CyclicDE never thrown from DPC if a parent container exists

          Details

          • Type: Bug Bug
          • Status: Resolved Resolved
          • Priority: Critical Critical
          • Resolution: Fixed
          • Affects Version/s: 1.2-RC1
          • Fix Version/s: 1.2
          • Component/s: PicoContainer (Java)
          • Labels:
            None
          • Testcase included:
            yes
          • Number of attachments :
            1

            Description

            If a DPC has a parent container, and component instantiation would cause a CyclicDE to be thrown, it is never thrown, and the component received 'null' instead.

            Attached is a patch to fix this, but it breaks two tests:

            DPCLifeCycleTestCase.testMaliciousComponentCannotExistInAChildContainerAndSeeAnyElementOfContainerHierarchy: It only returns "<One<Two"

            DPCTestCase.testUpDownDependenciesCannotBeFollowed: It can't instantiate A because of unsatisfied dependencies

            from what I can tell, the test behavior with this patch is still "ok", but would like others to review prior to commit.

              Activity

              peter royal made changes -
              Field Original Value New Value
              Summary CyclicDE never throw from DPC if a parent container exists CyclicDE never thrown from DPC if a parent container exists
              Hide
              Permalink
              Jörg Schaible added a comment -

              Well, if it breaks those unit tests, it is not good. Personally I fear it is a consequence of PICO-165 ...

              Show
              Jörg Schaible added a comment - Well, if it breaks those unit tests, it is not good. Personally I fear it is a consequence of PICO-165 ...
              Hide
              Permalink
              peter royal added a comment -

              It is a consequence of PICO-165

              If you could still take a peek though, because from what I could tell, the "spirit" of those tests is still honored, its just manifested in a new way. I could be wrong though

              Show
              peter royal added a comment - It is a consequence of PICO-165 If you could still take a peek though, because from what I could tell, the "spirit" of those tests is still honored, its just manifested in a new way. I could be wrong though
              Hide
              Permalink
              Jörg Schaible added a comment -

              Applied. You were right, Pete. Looking at the modified test cases UDE would have been better before also. Aslak, can you make a test against the latest SNAPSHOT?

              Show
              Jörg Schaible added a comment - Applied. You were right, Pete. Looking at the modified test cases UDE would have been better before also. Aslak, can you make a test against the latest SNAPSHOT?
              Jörg Schaible made changes -
              Assignee Aslak Hellesoy [ rinkrank ]
              Resolution Fixed [ 1 ]
              Status Open [ 1 ] Resolved [ 5 ]
              Hide
              Permalink
              Aslak Hellesøy added a comment -

              I noticed the exact same bug about a month ago - null getting injected instead of a CDE thrown. I'll comment here when/if we try to upgrade to the latest snapshot.

              Show
              Aslak Hellesøy added a comment - I noticed the exact same bug about a month ago - null getting injected instead of a CDE thrown. I'll comment here when/if we try to upgrade to the latest snapshot.

                People

                • Assignee:
                  Aslak Hellesøy
                  Reporter:
                  peter royal
                • Votes:
                  0 Vote for this issue
                  Watchers:
                  0 Start watching this issue

                  Dates

                  • Created:
                    Updated:
                    Resolved: