Does anyone know if it's ok to "implement" interface methods with abstract methods in the class that implements the interface? These methods will be overriden later in child classes. I mean, I could leave these methods as virtual in the class that implements the interface, and leave the method bodies empty.
Does anyone know if it's ok to "implement" interface methods with abstract methods in the class that implements the interface? These methods will be overriden later in child classes. I mean, I could leave these methods as virtual in the class that implements the interface, and leave the method bodies empty.
The compiler didn't complain.
The compiler didn't complain.
Stefan Glienke I understand your reasoning about not reporting (and it saddens me) but it could help others if another developer comes across the same (similar) issue, having it documented could save them time - fixed or not (if it isn't fixed, I am sure some choice words will be spoken about Emb but that is a different discussion).
ReplyDeleteIssues that will never be fixed +1
ReplyDeletehttps://quality.embarcadero.com/browse/RSP-18108
Stefan Glienke I've added the following comment to the QA report:
ReplyDeleteI also noted that it is even worse on multi-thread process.
In fact, the "xchg [esp],eax" opcode is a thread-safe instruction which implies some synchronization of all CPU cores cache (in older revisions of CPUs, it was even a full execution lock of all cores).
So it will flush all cores execution, even if there is no thread safety needed in this VMT resolution context.
So +1 to replace the code generation by the proposed jmp-based pattern.