This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Fix bootstrap with --disable-shared and linker plugin


* Jan Hubicka wrote on Sun, Jan 09, 2011 at 06:59:40PM CET:
> > 
> > Sure.  The question is whether --enable-lto should override
> > --disable-shared or vice versa.  I agree it is somewhat of a bikeshed
> > question, but it may be prudent to at least document the intended
> > semantics.
> 
> Still the decision needs to be done at toplevel Makefile.  We might opt --disable-shared
> to disable lto-plugin too, but it seems that users get this often wrong.

OK, that seems a good reason then.

> > > Toplevel configure already contains tests whether the plugin makes sense for
> > > host (enabling it for ELF and cygwin targets only) Is there easy test to see if
> > > shared libs can be built we can put into toplevel configure?
> > 
> > Well, you can add AC_PROG_LIBTOOL to configure.ac and test
> > $enable_shared = no afterwards, but that's not what I'd call an "easy
> > and lightweight test".  ;-)  I'm also not convinced it's a good idea to
> > introduce at this point in development.
> 
> Hmm, heavyweight indeed, but what are the risks of this solution?

Not too high I think.  But placement is important, toplevel configure.ac
and Libtool macros will both try to set LD, AR, DLLTOOL, NM, RANLIB,
STRIP, OBJDUMP, and the configure.ac settings need to take precedence.

Cheers,
Ralf


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