X-D
On Sep 8, 6:21 pm, William Stein <wst...@gmail.com> wrote:
> On Tue, Sep 8, 2009 at 9:15 AM,
>
>
>
>
>
> Juanjo<juanjose.garciarip...@googlemail.com> wrote:
>
> > On Sep 8, 2:36 pm, David Kirkby <david.kir...@onetel.net> wrote:
> >> I will be impossible to convince!!!
> >> I have no problem accepting the *compiler* needs -m64. That option
> >> tells the *compiler* to create 64-bit object files. But once the
> >> linker is asked to link the object files, it should be able to work
> >> out whether the executable to be created is 32-bit or 64-bit based on
> >> the type of object files it is being asked to link.
>
> > Well, I am amazed that you fail to refuse clear evidence. Run former
> > ECL's autoconf with CFLAGS=-m64 and LDFLAGS=. GCC is then used without
> > -m64 to link with libmpir and it fails to work with that libmpir you
> > built in 64 bits mode. Use CFLAGS=-m64 and LDFLAGS=-m64, and now GCC
> > gets also the -m64 flag because it is used as a linker and the
> > configuration works. Similar problems happened during ECL's build. I
> > really do not have a problem if you fail to admit that this happens.
> > ECL now inserts the -m64 flag whenever it invokes gcc for 64 bits
> > binaries -- compiling AND linking -- but you may face similar problems
> > with other libraries.
>
> > You also have to understand that -m64 passed to GCC may give more
> > information than the linker needs. For instance, GCC may have TWO
> > different runtime libraries, one for 32 bits and another one for 64
> > bits. If you do not tell GCC that you need 64-bits binaries, problems
> > will arise. So it is not only the linker that matters when considering
> > LDFLAGS.
>
> Pythagoras: "The Earth is round."
>
> William Carpenter: "Can you show me where it is documented that the
> Earth is round?"
>
> Pythagoras: "Well, I am amazed that you fail to refuse clear evidence."
>
> William Carpenter: " I will be impossible to convince!!!"
>
> -- William
--~--~---------~--~----~------------~-------~--~----~
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel-unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://www.sagemath.org
-~----------~----~----~----~------~----~------~--~---
