[PATCH] D22048: [analyzer] Suppress false positives in std::shared_ptr

[Devin Coughlin via cfe-commits]

dcoughlin added inline comments.

================
Comment at: test/Analysis/Inputs/system-header-simulator-cxx.h:349
@@ +348,3 @@
+    // No warning is expected as we are suppressing warning coming
+    // out of std::basic_string.
+    int z = 0;
----------------
NoQ wrote:
> You mean std::shared_ptr here?
> 
> Also, perhaps it's better to move this example into  separate fake method; if this header emulator is used in another test, wouldn't the author of that test be surprised to see that operator=() is modeled that way?
Yes, this is copy pasta!

================
Comment at: test/Analysis/Inputs/system-header-simulator-cxx.h:349
@@ +348,3 @@
+    // No warning is expected as we are suppressing warning coming
+    // out of std::basic_string.
+    int z = 0;
----------------
dcoughlin wrote:
> NoQ wrote:
> > You mean std::shared_ptr here?
> > 
> > Also, perhaps it's better to move this example into  separate fake method; if this header emulator is used in another test, wouldn't the author of that test be surprised to see that operator=() is modeled that way?
> Yes, this is copy pasta!
> Also, perhaps it's better to move this example into separate fake method; if this header emulator is used in another test, wouldn't the author of that test be surprised to see that operator=() is modeled that way?

We don't really model much in this this fake header, but I could move all the truly gross suppression stuff (basic string, _independent_bits_engine, shared_ptr) into a separate header and name it something that clearly indicates other tests shouldn't use it.

I'll update the patch.


http://reviews.llvm.org/D22048