anthonyfieroni added inline comments. INLINE COMMENTS
> dfaure wrote in kcompositejob.cpp:102 > No no no this is all wrong, please don't change the logic here. > After a subjob is done, we might want to start another one. > Most KIO jobs work like that. Yeah, you are right, but we ever emit on error? It's finish whole composite job (deleteLater will be called) > kcompositejob.cpp:68 > + job->setParent(nullptr); > + job->disconnect(this); > + return true; I rethink it, sorry. Let's remove only own connections. disconnect(job, &KJob::result, this, &KCompositeJob::slotResult); disconnect(job, &KJob::infoMessage, this, &KCompositeJob::slotInfoMessage); in two places. REPOSITORY R244 KCoreAddons REVISION DETAIL https://phabricator.kde.org/D9004 To: jtamate, #frameworks, dfaure, anthonyfieroni Cc: anthonyfieroni