Re: [Patch] Small refactor on <regex> _State<>

[Jonathan Wakely <jwakely@redhat.com>]

On 29/07/15 21:55 -0700, Tim Shen wrote:
>-      typedef _Matcher<typename _TraitsT::char_type> _MatcherT;
>+      typedef _Matcher<_Char_type> _MatcherT;
>+      static_assert(sizeof(_MatcherT) == sizeof(_Matcher<char>),
>+		    "The aussmption std::function<bool(char)> has "
>+		    "the same size as std::function<bool(T)> is violated");
>+      static_assert(alignof(_MatcherT) == alignof(_Matcher<char>),
>+		    "The aussmption std::function<bool(char)> has "
>+		    "the same alignment as std::function<bool(T)> is violated");

Good idea adding these assertions, but "aussmption" :-)

In accordance with https://gcc.gnu.org/bugzilla/show_bug.cgi?id=41759
we should avoid negative wording ("X is not Y" or "X does not meet Y")
in static assertions, so maybe something like:

      static_assert(alignof(_MatcherT) == alignof(_Matcher<char>),
		    "std::function<bool(T)> has the same size as "
		    "std::function<bool(char)>");

or:

      static_assert(alignof(_MatcherT) == alignof(_Matcher<char>),
		    "std::function<bool(T)> must have the same size as "
		    "std::function<bool(char)>");


>-      _MatcherT      _M_matches;        // for _S_opcode_match
>+      _MatcherT&
>+      _M_get_matcher()
>+      { return *reinterpret_cast<_MatcherT*>(this->_M_matcher_storage._M_addr()); }
>
>-      explicit _State(_Opcode __opcode) : _State_base(__opcode) { }
>+      const _MatcherT&
>+      _M_get_matcher() const
>+      { return *reinterpret_cast<const _MatcherT*>(this->_M_matcher_storage._M_addr()); }

These can use static_cast, because _M_addr() returns void*

OK for trunk with those two tweaks, thanks!