[Mpi-forum] C++ types inaccessible after #281

Jeff Hammond jhammond at alcf.anl.gov
Tue Jun 26 10:17:02 CDT 2012

Interesting, because
http://www.cplusplus.com/reference/std/complex/complex/, which I
recognize is not the C++ standard, just says that std::complex is
specialized for those types, not restricted to them.  Clang was happy
to compile std::complex<int> and related for me just now, so we might
want to worry about users doing this since we all know MPI users don't
read the MPI standard so why should we expect MPI-C++ users to read
both the MPI and C++ standards?

JeffS was not necessarily suggesting we replace the C++ standard with
a human being.  He was suggesting someone write the proposal for this
feature.  I don't see how a standards document is going to produce
draft text for the MPI standard.  Do you? :-)


On Tue, Jun 26, 2012 at 10:05 AM, Jed Brown <jedbrown at mcs.anl.gov> wrote:
> On Tue, Jun 26, 2012 at 6:33 AM, Jeff Hammond <jhammond at alcf.anl.gov> wrote:
>> So one issue to consider here is that std::complex is a template, not
>> a single type.  One can presumably stick lots of types inside the
>> brackets.
> Good thing the C++ standards committee nipped that one in the bud:
> 26.4
> The effect of instantiating the template complex for any type other than
> float, double, or long double is unspecified. The specializations
> complex<float>, complex<double>, and complex<long double> are literal types
> (3.9).
>> So the MPI Forum needs to figure out which of the existing C99 types
>> that will work with this.  Do we support std::complex<long double>?
>> How about std::complex<uint64_t>?
> No
>> My reference is http://www.cplusplus.com/reference/std/complex/complex/,
>> btw.
> As a standards guy, why not use a standard (or a final draft, which you can
> have for free)?
> http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2011/n3242.pdf
> _______________________________________________
> mpi-forum mailing list
> mpi-forum at lists.mpi-forum.org
> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum

Jeff Hammond
Argonne Leadership Computing Facility
University of Chicago Computation Institute
jhammond at alcf.anl.gov / (630) 252-5381

More information about the mpi-forum mailing list