[PATCH] [Review request][analyzer]Information about special handling for a particular region/symbol.

Антон Ярцев anton.yartsev at gmail.com
Tue Sep 24 16:14:58 PDT 2013

  Took away default NULL initialization.
  OK to commit?

Comment at: include/clang/StaticAnalyzer/Core/CheckerManager.h:377
@@ -374,3 +376,3 @@
                               PointerEscapeKind Kind,
-                              bool IsConst = false);
+                              RegionAndSymHandlingTraits *HTraits = NULL);
Anna Zaks wrote:
> Антон Ярцев wrote:
> > Anna Zaks wrote:
> > > When does it make sense for traits to be missing? (Why the NULL initialization? Also, I think it should be '0', not "NULL")
> > Currently pointer escape on bind do not deal with traits ( ExprEngine.cpp, processPointerEscapedOnBind() ), NULL initialization just allows to leave the code of processPointerEscapedOnBind() unchanged. This is the only reason for NULL initialization. Do you think NULL initialization should be removed?
> > 
> > Changed "NULL" to "0". Searched LLVM code standard for "0" vs "NULL", found nothing. What is wrong with "NULL"? :)
> > 
> Is that expected to change? Maybe we should just pass NULL at that one call site, instead of using default initialization.
> There are probably threads discussing which one to use "NULL" or "0" and I assume we came to a decision to use '0'. I can see arguments on both sides, one is more expressive, the other one is shorter and cannot be redefined...


More information about the cfe-commits mailing list