Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

C++11 support (esp concurrency)

20 views
Skip to first unread message

tortoise underscore ignoreme 74 at yahoo. nospam co.uk

unread,
Dec 6, 2011, 11:59:17 AM12/6/11
to
Hi,
I am looking into the feasability of adopting the subset of C++11
supported by gcc 4.7. I believe there is a strong case for doing so
from just the minor features and library improvements. However, I am
having a hard time putting together a convincing cost benefit analysis
(any help with this would be great).

On the side of major enhancements would be support for multi-
threading. The status page and manual give
only the surface detail

http://gcc.gnu.org/gcc-4.7/cxx0x_status.html
http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#status.iso.2011

It is clear that support is not 100% but it is not that clear what is
missing and how far away it might be.
Does anyone have more detailed information?

In particular I would like to know what core language features are
working on GNU/Linux and which are absent.
Which omissions are solely library problems?

Another thing that is unclear is the status of standard libraries.
Some may be listed as not supported that are available in boost.
Presumably these are awaiting someone deriving or re-implementing a
version under the GPL. The fact that an open source (though not GPL)
version is available will presumably make these lower priorities to
develop (regexp is a notable absence for example).
From a users perspective they can have a working implementation if
they choose to combine gcc and boost
(I see just::threads is being sold as a commercial option).

So to summarise what doesn't work but can be (relatively easily)
worked around
and what doesn't work that can't be worked around?

Regards,

Bruce.

Andrew Haley

unread,
Dec 8, 2011, 7:30:26 AM12/8/11
to
tortoise underscore ignoreme 74 at yahoo. nospam co.uk <torto...@yahoo.co.uk> wrote:

> I am looking into the feasability of adopting the subset of C++11
> supported by gcc 4.7. I believe there is a strong case for doing so
> from just the minor features and library improvements. However, I am
> having a hard time putting together a convincing cost benefit analysis
> (any help with this would be great).
>
> On the side of major enhancements would be support for multi-
> threading. The status page and manual give
> only the surface detail
>
> http://gcc.gnu.org/gcc-4.7/cxx0x_status.html
> http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#status.iso.2011
>
> It is clear that support is not 100% but it is not that clear what is
> missing and how far away it might be.
> Does anyone have more detailed information?
>
> In particular I would like to know what core language features are
> working on GNU/Linux and which are absent.

That depends on the distro. Basically, everything marked "Yes" will
work with the latest gcc/libc sources.

> Which omissions are solely library problems?
>
> Another thing that is unclear is the status of standard libraries.
> Some may be listed as not supported that are available in boost.
> Presumably these are awaiting someone deriving or re-implementing a
> version under the GPL. The fact that an open source (though not GPL)
> version is available will presumably make these lower priorities to
> develop (regexp is a notable absence for example).
> From a users perspective they can have a working implementation if
> they choose to combine gcc and boost
> (I see just::threads is being sold as a commercial option).

Indeed.

> So to summarise what doesn't work but can be (relatively easily)
> worked around and what doesn't work that can't be worked around?

That's a huge question. It would take quite a lot of research to find
out. I doubt that any single person can give a full answer.

Andrew.
0 new messages