Sytem information: - LFS-based Linux, with kernel 2.6.15 (with some genpatches) GCC configuring --------------- ./gcc-3.4.3/configure --prefix=/usr \ --libexecdir=/usr/lib --enable-shared \ --enable-threads=posix --enable-__cxa_atexit \ --enable-clocale=gnu --enable-languages=c,c++ GCC Compiling: ------------- make bug description: ----------------- We compiled the attached source code using the command: > g++ --save-temps build_and_destroy.cc -o bug Running ./bug produced the following output: A A copy A ~A A.m:1 ~A ~A As you can see, returning the object in the function F didn't call the copy constructor. We expect the following output: A A copy A A copy ~A ~A A.m:2 ~A ~A
Created attachment 12088 [details] The original source code Bug related file
Created attachment 12089 [details] The preprocessed source (*.ii) file Bug related file
this is not a bug. the c++ standard allows RVO optimization. see $12.8/15 for more details.
Never thought the C++ standard allows ignoring the code written in a dtor. What if the dtor does something important (like in the example - it doesn't just copy the contents)?
I meant the copy ctor, not the dtor. Anyhow, looking at it again, it makes some sense.
(In reply to comment #5) > I meant the copy ctor, not the dtor. > Anyhow, looking at it again, it makes some sense. "When certain criteria are met, an implementation is allowed to omit the copy construction of a class object, even if the copy constructor and/or destructor for the object have side effects. In such cases, the implementation treats the source and target of the omitted copy operation as simply two different ways of referring to the same object" ps). you can disable such optimization with -fno-elide-contructors
This is not a bug as mentioned by comment #6.