[cfe-dev] Replacing the analyzer's CallOrObjCMessage with something more general

[Jordan Rose]

On Jun 28, 2012, at 15:55 , Tom Care <tom.care at uqconnect.edu.au> wrote:

> Looks great, this makes my C++ constructor inlining significantly less hacky. I'll have to integrate and see how it works with my branch, but I'm hoping to be able clean it up enough to contribute new/delete handling very soon, including MallocChecker support.
> 
> Tom

That would be awesome. Please take note of http://llvm.org/bugs/show_bug.cgi?id=12014 if you work on new/delete, though -- the upshot is that the visit order for the sub-operations in a new-expression is pretty unusual.

For my part, I'm planning to work on enabling destructors once this gets in; turning on constructors without destructors pretty much results in false positives everywhere. Once we have ctors, dtors, new, and delete, we'll probably be able to turn on analysis for a huge portion of C++ code.

Jordan