Re: debug container mutex association

[Jonathan Wakely <jwakely@redhat.com>]

On 28/09/16 21:30 +0200, François Dumont wrote:
>On 27/09/2016 12:32, Jonathan Wakely wrote:
>>Index: include/debug/safe_base.h
>>>===================================================================
>>>--- include/debug/safe_base.h    (revision 240509)
>>>+++ include/debug/safe_base.h    (working copy)
>>>@@ -121,11 +121,11 @@
>>>    void
>>>    _M_detach();
>>>
>>>+  public:
>>>    /** Likewise, but not thread-safe. */
>>>    void
>>>    _M_detach_single() throw ();
>>>
>>>-  public:
>>>    /** Determines if we are attached to the given sequence. */
>>>    bool
>>>    _M_attached_to(const _Safe_sequence_base* __seq) const
>>
>>
>>Would this be a smaller change, that doesn't make the member
>>accessible to all code?
>>
>>--- a/libstdc++-v3/include/debug/safe_base.h
>>+++ b/libstdc++-v3/include/debug/safe_base.h
>>@@ -50,6 +50,7 @@ namespace __gnu_debug
>>  class _Safe_iterator_base
>>  {
>>    friend class _Safe_sequence_base;
>>+    template<typename> friend class _Safe_sequence;
>>
>>  public:
>>    /** The sequence this iterator references; may be NULL to indicate
>>.
>>
>I am not a great fan of friend class. As long as it was friend 
>declaration between iterator and sequence base types it was ok. Now 
>that we need to make a template class friend I consider that it is too 
>much friendship and prefer to make the necessary method public.

OK.

>But if you think otherwise just tell me and I will use your approach.

Making it public is fine. Please commit your original patch, thanks!