Re: C++0x issues

[Paolo Carlini <paolo dot carlini at oracle dot com>]

Hi,

> I have run the parallel testsuite with and without C++0x, and
> everything seems fine. Nothing fails which doesn't fail either with
> C++0x or parallel in isolation.

Excellent, thanks.

> On a related note, we should probably start running the testsuite in
> C++0x mode more regularly, and somehow mark tests which won't pass in
> C++0x. I'm not sure how hard that is to do.
>
> Just running the testsuite in C++0x threw up a few things, in
> particular it bought Library issue 1133 to my attention, which
> basically means that all C++03 code which calls list::splice won't
> work in C++0x.

Yes, this is a known issue, I believe you can find traces of it either
in the mailing lists or in Bugzilla (normally we don't keep open PRs
which simply amount to issues with the current wp or CD, provided a DR
exists)

> Looking at issue 1133, while there is no Proposed Resolution yet, I
> intend to test a patch for option (A) there, giving both a non-const
> lvalue and rvalue overload, which seems the only way to keep C++03
> code working, while also allowing code like l.splice(l.end(),
> function_returning_list()); to work.

Thanks for the help, the same problem affects forward_list too if I
remember correctly, better also checking the other mambers... Anyway, in
Santa Cruz I'll make sure people are by and large on the same page about
this, then we'll be able to commit the changes in time for 4.5.0, seems
a good idea.

Thanks,
Paolo