libc++: First cut at <dynarray>

[Howard Hinnant]

On Sep 13, 2013, at 12:13 AM, Chandler Carruth <chandlerc at google.com> wrote:

> On Thu, Sep 12, 2013 at 7:55 PM, Howard Hinnant <howard.hinnant at gmail.com> wrote:
> 
>> Sorry Marshall.  Great job, but hold off on the commit.  I think we need to sort this out.
> 
> Jumping up a few layers of context...
> 
> I'm actually fine checking in dynarray with a heap-only implementation. If anything, so that we can demonstrate more fully that yes, we implemented everything, and shockingly enough found problems. If the problems persuade the committee to reconsider dynarray for C++14, wonderful. But it doesn't seem unreasonable to track the draft as it progresses...

Ok, thanks for your comments on this Chandler.

This looks like a mess to me.

But let's get Marshall's work checked in.  At the very least it will give the clang team a little easier access to this work for experimentation.

This is the first time I've committed beta software to libc++.  Up until now libc++ tip-of-trunk is always in a shippable state.  In this case things are mitigated by the #if _LIBCPP_STD_VER > 11 protection.  But I don't want to make a habit of checking in half-baked projects.

My goal is to get dynarray into a shippable state, or remove it, asap.  I consider it not-shippable as long as it has no stack allocation support.

Sorry for jerking you around on this Marshall.  You've done a great job with this.  Please commit.

Howard