This is the mail archive of the libstdc++@sources.redhat.com mailing list for the libstdc++ project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: changing include/g++-v3 to something a bit more obvious


>>>>> "Phil" == Phil Edwards <pedwards@disaster.jaj.com> writes:

    Phil> Anyhow, if the compiler version is going to play a part in
    Phil> the directory naming, why not just put the tree under
    Phil> gcc-lib/platform/version?  The main obstacle to the
    Phil> compiler's -V option is that too many of the include files
    Phil> are outside of the -V-controlled tree in the first place.

    Phil> There's already an 'include' subdir; we could just make it
    Phil> 'include/g++' and let the pathname speak for itself.

The difficulty here is that, in theory, if we actually get the C++ ABI
stable (which it's supposed to be after 3.0), you might want to use
the same old library with a brand-new compiler.  (Why?  Because the
library ABI is bound to change; it's not fully done yet.  So, if you
want to use a new compiler with better optimization, etc., but still
produce link-compatible stuff you might stick with the old library.)

I'm not sure exactly what that means; it's just a data-point.

I agree that making -V actually work would be terrfic, so I think on
balance I like Phil's suggestion.  It's going to need discussion on
the GCC lists, and possibly even with the SC, though; I can't decide
this unilaterally.

--
Mark Mitchell                   mark@codesourcery.com
CodeSourcery, LLC               http://www.codesourcery.com

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]