[llvm] r187018 - Replace with a "null" RAUW with an assert since we'd actually like

[David Blaikie]

On Tue, Jul 23, 2013 at 6:51 PM, Eric Christopher <echristo at gmail.com> wrote:
> Author: echristo
> Date: Tue Jul 23 20:51:12 2013
> New Revision: 187018
>
> URL: http://llvm.org/viewvc/llvm-project?rev=187018&view=rev
> Log:
> Replace with a "null" RAUW with an assert since we'd actually like
> to assume we're replacing. Clarify comments.
>
> Modified:
>     llvm/trunk/lib/IR/DebugInfo.cpp
>
> Modified: llvm/trunk/lib/IR/DebugInfo.cpp
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/IR/DebugInfo.cpp?rev=187018&r1=187017&r2=187018&view=diff
> ==============================================================================
> --- llvm/trunk/lib/IR/DebugInfo.cpp (original)
> +++ llvm/trunk/lib/IR/DebugInfo.cpp Tue Jul 23 20:51:12 2013
> @@ -327,11 +327,11 @@ unsigned DIArray::getNumElements() const
>    return DbgNode->getNumOperands();
>  }
>
> -/// replaceAllUsesWith - Replace all uses of debug info referenced by
> -/// this descriptor.
> +/// replaceAllUsesWith - Replace all uses of the MDNode used by this
> +/// type with the one in the passed descriptor.
>  void DIType::replaceAllUsesWith(DIDescriptor &D) {
> -  if (!DbgNode)
> -    return;
> +
> +  assert(DbgNode && "Trying to replace an unverified type!");
>
>    // Since we use a TrackingVH for the node, its easy for clients to manufacture
>    // legitimate situations where they want to replaceAllUsesWith() on something
> @@ -347,11 +347,11 @@ void DIType::replaceAllUsesWith(DIDescri
>    }
>  }
>
> -/// replaceAllUsesWith - Replace all uses of debug info referenced by
> -/// this descriptor.
> +/// replaceAllUsesWith - Replace all uses of the MDNode used by this
> +/// type with the one in D.
>  void DIType::replaceAllUsesWith(MDNode *D) {
> -  if (!DbgNode)
> -    return;
> +
> +  assert(DbgNode && "Trying to replace an unverified type!");

Is this (& above) the right assertion text? Since we don't actually do
the verification/isFoo stuff in the ctors - the null case just happens
when we construct a DIFoo from a null DbgNode - probably because the
node was just null (ie: we didn't build with debug info), I think...
perhaps I'm wrong, maybe they become null for other reasons?

>
>    // Since we use a TrackingVH for the node, its easy for clients to manufacture
>    // legitimate situations where they want to replaceAllUsesWith() on something
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits