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

[Jonathan Wakely]

On 29/07/15 01:43 -0700, Tim Shen wrote:
>On Wed, Jul 29, 2015 at 1:32 AM, Jonathan Wakely <jwakely@redhat.com> wrote:
>> Apologies, you have a user-declared move constructor, so assignment is
>> already deleted. It wouldn't hurt to make that explicit though:
>
>I'm just bad at memorizing when they are implicitly
>declared/defined/deleted and fully unware of the deleted copy
>assignment :)
>
>It's better to make it explicit, since I don't want to be clever on
>pretending knowing all ctor/assigment rules :)

Yes, I think it's better to make it explicit.

>>  _State& operator=(const _State&) = delete;
>>
>> So it's just the alignment issue that I'm concerned about now.
>>
>
>Will alignof(std::function<bool(C)>) be all the same across
>instantiations, including user-defined C, now and in the future?

All member variables of std::function<> are in the _Function_base base
class, which is not a template, so it has the same size and alignment
for all instantiations.

>I created template parameter _Matcher_size (which should be
>__matcher_size) because I don't want to make this assumption on
>sizeof(). If they (both alignment and size) are expected to be the
>same, we can remove that template parameter (and all indentation
>changes!);

Yes, that makes sense. See the code in <ext/aligned_buffer.h> for how
to set the alignment of the buffer appropriately. You can use the size
and alignment of std::function<bool(char)> even though it will
sometimes be a different std::function specialization.