The only dangerous thing about mixing C and C++ I can think of is PETSc's reliance on the specific struct layout when downcasting (e.g., to PetscObject). What does the C++ standard say about this?
Dmitry. On Mon, Apr 4, 2011 at 8:48 PM, Satish Balay <balay at mcs.anl.gov> wrote: > There is clang and clang++ - but still they insist on throwing this > warning. > > When I invoke clang++ - I want c++. So I'm not sure whats so ambiguous > that they should throw this warning. > > And what does this depricate mean? That 'clang++ foo.c' will create C > code? > > Then why have both clang and clang++? > > satish > > On Mon, 4 Apr 2011, Barry Smith wrote: > > > > > configure.log ???? > > > > What about passing the compiler option -x C++? Or maybe you are > already doing that but I would't know because you didn't send configure.log > > > > It's bad enough when our ignorant users don't send necessary > configure.log files but you? > > > > Barry > > > > On Apr 4, 2011, at 8:33 PM, Sean Farley wrote: > > > > > When specifying --with-clanguage=C++ (as required by sieve) all C code > (any .c file) is still compiled with the C++ compiler. This behavior is > deprecated as clang will warn you: > > > > > > clang: warning: treating 'c' input as 'c++' when in C++ mode, this > behavior is deprecated > > > > > > C is not a subset of C++ and treats structs very differently than C++. > This should really be handled correctly but I can't read the conf/rules file > no more than I can read Greek. > > > > > > Sean > > > <make.log> > > > > > > -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20110404/85d8f412/attachment.html>
