Re: Mothballing C11 atomic work for now.

[Jeff Hammond <jeff dot science at gmail dot com>]

If C11 atomics are not going into 4.9, then comments made to reject
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58016 no longer hold and I
would ask that the resolution of both it and
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53769 be reconsidered.

Jeff

On Fri, Sep 27, 2013 at 2:10 PM, Andrew MacLeod <amacleod@redhat.com> wrote:
> I don't have the time to finish pushing through the C11 atomic work for this
> release. Much of the remaining parts are in the parser, which I know very
> little about, and I won't be able to do a sufficient job in the time
> remaining, so I am switching my focus to the interface work and getting the
> header files re-factored before we end stage 1.
>
> I have put all the work to date in a branch 'C11-atomic' which is based off
> of trunk on Sept 25th.  I have detailed the current status as well as what
> else needs doing on the C11 atomic wiki page
> http://gcc.gnu.org/wiki/Atomic/C11
>
> I have also uploaded the specific patches that have been applied to the
> branch, along with their revision numbers.  they are also on that wiki page.
>
> The patches are very similar to what I posted here
> http://gcc.gnu.org/ml/gcc-patches/2013-08/msg00420.html
> I addressed jsm's basic comments, but did not address the larger issues of
> warnings/errors and converting lvalues into rvals, and other front end
> issues.  I also removed the places where i tried to treat the atomic
> qualifier like it was volatile.. I think that was wrong and was masking
> other issues.
>
> If this work is important to someone else, you are welcome to pick it up.
> My parser expertise is minimal, and most of the remaining work is in that
> part of the compiler.   The facilities are provided already to do the
> expansion of atomic variable into the appropriate sequences, they just need
> to be called from the right places in the parser.
>
> I may well get back to this next spring for the next release, but for now I
> am mothballing it until I have the time to learn that parts I need to learn
> to finish it.
>
> Andrew



-- 
Jeff Hammond
jeff.science@gmail.com