On Fri, Nov 20, 2009 at 4:23 AM, rahul <[email protected]> wrote:
> I was trying to compile chromium today on Linux and I got this error.
> This error occurred because of -Werror CFLAG set in the Makefile. I
> remedied by writing a one-liner to delete all occurrences of -Werror
> from CFLAGS in all Makefiles.
>
> However, this error intrigued me. I saw through the code and the
> classes in question have declared virtual methods but haven't declared
> a destructor. In that case, the compiler generates the destructor.
>
>
> Base *b = new Sub();
> delete b;
>
> To the best of my knowledge, if Base is the class which defines
> virtual methods but non-virtual destructor, delete can cause a memory
> leak if Sub happens to be taking more memory than Base. I think  that
> the destructor for a class with virtual methods has either to be
> "public virtual" or "protected non-virtual" for the proper freeing of
> memory.
>
>
> Would someone help me clarify this?
>
> I know there might be some scenarios where this can be done
> intentionally but looks like there are more arguments for adding a
> virtual destructor than not adding one. I see that this is done for
> the Delegates. Also, if this is one of the intended special cases, can
> someone explain how to compile chromium  then? Deleting -Werror from
> CFLAGS is an ugly hack at best.

Your analysis is correct.  However, a virtual destructor is not needed
in the case where you never delete through the Base*.  It turns out
for our codebase that is very common (due to lots of "observer"-like
patterns), so we decided to not rely on this compiler warning.  It's
only caused horrible bugs once or twice!  :)

The correct fix would be to disable this particular warning in
build/common.gypi.
However, I'm puzzled why nobody else has encountered the problem
you're mentioning.
Can you provide some details about your compiler, system configuration, etc.?

-- 
Chromium Developers mailing list: [email protected] 
View archives, change email options, or unsubscribe: 
    http://groups.google.com/group/chromium-dev

Reply via email to