[cfe-dev] C++ Language Support Library

[Doug Gregor]

On Mon, Nov 10, 2008 at 7:23 PM, Howard Hinnant <hhinnant at apple.com> wrote:
> On Nov 10, 2008, at 6:54 PM, Chris Lattner wrote:
>
>> On Nov 10, 2008, at 3:00 PM, Doug Gregor wrote:
>>>> For a random linux distro, for example, it would be fine to compile
>>>> libstdc++ with GCC, and clang should be able to work with it.
>>>>
>>>> I think that retaining compatibility with libstdc++ would be a very
>>>> worthwhile goal (as is supporting the apache library, stlport, and/
>>>> or
>>>> whatever other ones exist, where reasonable).  What is the cost of
>>>> doing
>>>> this?  Does it use crazy GCC extensions that we don't want to
>>>> implement?
>>>
>>> libstdc++ aims to be pretty standard-conforming, and they avoid most
>>> crazy GCC extensions. As Sebastian noted, in GCC 4.3 and newer are
>>> using some C++0x features in their TR1 implementation. The biggest
>>> feature there is variadic templates, which---while not trivial---
>>> isn't
>>> terribly hard to implement if the template system itself is designed
>>> well.
>>
>> Ok! I agree with Sebastian that libstdc++ support will be critical for
>> clang C++ adoption.  While libstdc++ *is* a moving target, picking a
>> very specific one (say, GCC 4.2's) should be useful enough in the
>> short term, and doesn't require an '0x support to handle.  I picked
>> 4.2 specifically because it is used on the mac, but any other
>> reasonable one would be good.
>
> For libstdc++ 4.2 we will need support for "strong using" attributes
> with the following extension syntax:
>
> namespace std
> {
>   namespace _6 { }
>   using namespace _6 __attribute__ ((strong));
> }
>
> This is a C++0X feature which will have C++0X syntax more like:
>
> namespace std
> {
>   inline namespace _6
>   {
>       /*... declarations and definitions*/
>   }
> }
>
> Because of the structure of gcc 4.2, I believe both syntaxes will
> ultimately need to be supported, but initially, only the gcc 4.2
> syntax.  I.e. I don't believe the library can transform one syntax to
> the other with macros (but I could be wrong).

For what it's worth, the rewrite they did to change from "strong
using" to "inline namespaces" was quite trivial:

  http://gcc.gnu.org/viewcvs/trunk/libstdc%2B%2B-v3/include/bits/c%2B%2Bconfig?r1=131982&r2=132611&diff_format=h

So it shouldn't be too painful to support both features.

  - Doug