Re: [wwwdocs] Entry for PR 18160 in changes.html

[Gerald Pfeifer <gerald at pfeifer dot com>]

On Sat, 13 Nov 2004, Adam Nemet wrote:
> How about this?  It still passes the validator.

Thanks, this looks fine!

There were CVS conflicts when applying the second hunk, and a minor
indentation issue in the <pre> environment, so I went ahead and committed
the following, slightly updated patch.

Gerald

Index: changes.html
===================================================================
RCS file: /cvs/gcc/wwwdocs/htdocs/gcc-4.0/changes.html,v
retrieving revision 1.15
diff -u -3 -p -r1.15 changes.html
--- changes.html	2 Nov 2004 10:48:55 -0000	1.15
+++ changes.html	14 Nov 2004 21:29:27 -0000
@@ -100,6 +100,9 @@
         deprecated in 3.4, has been removed.</li>
     <li><code>#pragma pack()</code> semantics have been brought closer to
         those used by other compilers. This also applies to C++.</li>
+    <li>Taking the address of a variable with <code>register</code>
+        storage is invalid in C.  GCC now issues an error instead of a
+        warning.</li>
   </ul>

 <h3>C++</h3>
@@ -144,6 +147,17 @@
         want to disable this by specifying
         <code>-fno-threadsafe-statics</code> for a small savings in code
         size.</li>
+
+    <li>Taking the address of an explicit register variable is no
+        longer supported.  Note that C++ allows taking the address of
+        variables with <code>register</code> storage so this will
+        continue to compile with a warning.  For example, assuming
+        that <code>r0</code> is a machine register:
+        <pre>register int foo asm ("r0");
+register int bar;
+&amp;foo; // error, no longer accepted
+&amp;bar; // OK, with a warning</pre></li>
+
     <li>G++ has an undocumented extension to virtual function
 	covariancy rules that allowed the overrider to return a type
 	that was implicitly convertable to the overridden function's