[clang] Add test for CWG600

[Endill]

Summary

P1787: CWG600 is resolved by explaining that accessibility affects naming a member in the sense of the ODR.
Wording: see changes to [class.access] p1 and p4.
Additional references: basic.def.odr/8: A function is odr-used if it is named by a potentially-evaluated expression or conversion.

[shafik]

clang/test/CXX/drs/dr6xx.cpp

18

Maybe add a comment above this saying something like:

// access control is applied after overload resolution
// [class.access.general]p4 "For an overload set, access control is applied only to the function selected by overload resolution."

[shafik]

Thank you for updating the DR statuses! This is much appreciated.

[Endill]

clang/test/CXX/drs/dr6xx.cpp

18

I tend to like the idea, but I wonder about general rule for adding such explanations. Currently DR tests contain very little of those.

If we're going to add explanations, we should also decide whether we're going to cite the standard, or paraphrase (and/or) explain intent. My concern is that both references to standard and citations could grow old relatively quickly, and we don't have any tools to help, at least yet.

[Endill]

clang/test/CXX/drs/dr6xx.cpp

18

@aaron.ballman What do you think?

[aaron.ballman]

clang/test/CXX/drs/dr6xx.cpp

18

We don't typically add a significant amount of comments to test files unless what is being tested needs some explanation. So, IMO, if the DR test case is "tricky" in some way and we're trying to demonstrate that we're testing a specific sentence or two from the standard, then I think a comment with standards wording is reasonable (please don't just cite [foo.bar]p12 though -- add the standards wording!). However, if the test is pretty straightforward, or the DR explains in more detail what's going on, then I don't think we need to add the comment (the references get stale rather quickly, even with stable names as in C++).

All that said, if you're on the fence about whether to add a comment or not, go ahead and add it (IMO).

[Endill]

clang/test/CXX/drs/dr6xx.cpp

18

I think that this one is the case when DR explains it better, because that's what it is about. Adding a comment seems more appropriate for something that's not given enough attention or accent in the whole context of a DR test, as opposed to test code itself.

@shafik Is it fine by you if I don't add any comment?

[shafik]

clang/test/CXX/drs/dr6xx.cpp

18

I would be satisfied with this comment alone, it would avoid the concern about wording and paragraph numbering shifting around in subsequent drafts:

// access control is applied after overload resolution

While the DR does explain the issue it is not obvious what the resolution is without spending some time digging, but at least with the comment you know what your looking for if you want to do the digging.

[Endill]

Add a comment per @shafik request

[aaron.ballman]

LGTM!

clang/test/CXX/drs/dr6xx.cpp

 // RUN: %clang_cc1 -std=c++98 %s -verify -fexceptions -fcxx-exceptions -pedantic-errors -fno-spell-checking
 // RUN: %clang_cc1 -std=c++11 %s -verify -fexceptions -fcxx-exceptions -pedantic-errors -fno-spell-checking
 // RUN: %clang_cc1 -std=c++14 %s -verify -fexceptions -fcxx-exceptions -pedantic-errors -fno-spell-checking
 // RUN: %clang_cc1 -std=c++17 %s -verify -fexceptions -fcxx-exceptions -pedantic-errors -fno-spell-checking
 // RUN: %clang_cc1 -std=c++20 %s -verify -fexceptions -fcxx-exceptions -pedantic-errors -fno-spell-checking
+// RUN: %clang_cc1 -std=c++2b %s -verify -fexceptions -fcxx-exceptions -pedantic-errors -fno-spell-checking
+
+namespace dr600 { // dr600: yes
+struct S {
+  void f(int);
+
+private:
+  void f(double); // expected-note {{declared private here}}
+};
+
+void g(S *sp) {
+  sp->f(2);
+  // access control is applied after overload resolution
+  sp->f(2.2); // expected-error {{is a private member}}
+}
+} // namespace dr600
 
 namespace std {
   struct type_info {};
   __extension__ typedef __SIZE_TYPE__ size_t;
 } // namespace std
 
 namespace dr601 { // dr601: yes
 #if __cplusplus >= 201103L

clang/www/cxx_dr_status.html

<tr id="599">
     <td>CD2</td>
     <td>Deleting a null function pointer</td>
     <td class="partial" align="center">Partial</td>
   </tr>
   <tr id="600">
     <td><a href="https://wg21.link/cwg600">600</a></td>
     <td>CD6</td>
     <td>Does access control apply to members or to names?</td>
-    <td class="none" align="center">Unknown</td>
+    <td class="full" align="center">Yes</td>
   </tr>
   <tr id="601">
     <td><a href="https://wg21.link/cwg601">601</a></td>
     <td>CD2</td>
     <td>Type of literals in preprocessing expressions</td>
     <td class="full" align="center">Yes</td>
   </tr>
   <tr id="602">