16732 - Need definition of inline functions before use in nsDST.cpp

Status: VERIFIED FIXED

(Core :: Layout, defect, P3)

Description

[tor]

Your recent change to layout/html/base/src/nsDST.cpp (1.12) caused a couple
of inline functions to be called before they were defined.  The native compiler
on my platform (Solaris) can't deal with this.  This simple patch moves the
definitions of nsDST::DestroyNode before nsDST::FreeTree:

Index: nsDST.cpp
===================================================================
RCS file: /cvsroot/mozilla/layout/html/base/src/nsDST.cpp,v
retrieving revision 1.13
diff -u -r1.13 nsDST.cpp
--- nsDST.cpp   1999/10/18 20:16:49     1.13
+++ nsDST.cpp   1999/10/18 22:53:14
@@ -305,6 +305,24 @@
   mArena->Release();
 }

+// Called by Remove() to destroy a node. Explicitly calls the destructor
+// and then asks the memory arena to free the memory
+inline void
+nsDST::DestroyNode(LeafNode* aLeafNode)
+{
+  aLeafNode->~LeafNode();     // call destructor
+  mArena->FreeNode(aLeafNode); // free memory
+}
+
+// Called by Remove() to destroy a node. Explicitly calls the destructor
+// and then asks the memory arena to free the memory
+inline void
+nsDST::DestroyNode(TwoNode* aTwoNode)
+{
+  aTwoNode->~TwoNode();      // call destructor
+  mArena->FreeNode(aTwoNode); // free memory
+}
+
 void
 nsDST::FreeTree(LeafNode* aNode)
 {
@@ -358,24 +376,6 @@
     aNode = ((TwoNode*)aNode)->mRight;
     goto keepLooping;
   }
-}
-
-// Called by Remove() to destroy a node. Explicitly calls the destructor
-// and then asks the memory arena to free the memory
-inline void
-nsDST::DestroyNode(LeafNode* aLeafNode)
-{
-  aLeafNode->~LeafNode();     // call destructor
-  mArena->FreeNode(aLeafNode); // free memory
-}
-
-// Called by Remove() to destroy a node. Explicitly calls the destructor
-// and then asks the memory arena to free the memory
-inline void
-nsDST::DestroyNode(TwoNode* aTwoNode)
-{
-  aTwoNode->~TwoNode();      // call destructor
-  mArena->FreeNode(aTwoNode); // free memory
 }

 nsDST::LeafNode*

Comment 1

[troy]

Where in the language does it say that inline functions need to be defined
before they're called?

Or is this a known restriction of the Solaris compiler?

Comment 2

[tor]

The compiler is also happy if you tell it that the function/method is going
to be inline and then call it before the definition.  Thus the following code
is acceptable, but if you remove the "inline" in the class definition the
compiler will report an error:
class X {
 inline int foobar(void);
 void baz(void);
};

void X::baz(void) { foobar(); }

inline int X::foobar(void) { return 0; }

The part of C++ specification relevant to this is probably 7.1.1.7
(dcl.stc part 7) - "The linkages implied by successive declarations
for a given entity shall agree.  That is, within a given scope, each
declaration declaring the same object or the same overloading of a
function name shall imply the same linkage."

Comment 3

[troy]

That doesn't really make sense. If it were true that the linkage of the
declaration and the definition must agree, then how come the compiler is happy
if the definitions of DestroyNode() come before FreeTree()? The linkages don't
agree in that case

Comment 4

[troy]

Moved the DestroyNode() functions before FreeTree() to make the Solaris compiler
happy

Comment 5

[Chris Petersen]

Based on Troy's remarks, marking as verified fixed.