Re: libstdc++/9776: -static-libgcc does not produce a static link to libstdc++, instead always dynamic

[Phil Edwards <phil at jaj dot com>]

The following reply was made to PR libstdc++/9776; it has been noted by GNATS.

From: Phil Edwards <phil at jaj dot com>
To: stew at tenbox dot com
Cc: gcc-gnats at gcc dot gnu dot org
Subject: Re: libstdc++/9776: -static-libgcc does not produce a static link to libstdc++, instead always dynamic
Date: Thu, 20 Feb 2003 11:44:39 -0500

 On Thu, Feb 20, 2003 at 04:30:56PM -0000, stew at tenbox dot com wrote:
 > 
 > I worry that maybe I am confused about what -static-libgcc does; maybe it
 > is not supposed to make the link to libstdc++ static.
 
 It isn't.  If you'll look at your ldd output, you'll see that no dependency
 on libgcc_s.so was recorded.  That's why the option is called -static-libgcc,
 not -static-libstdc++.
 
 > If not, then I
 > would suggest that such an option is very needed - it is not often
 > reasonable that the user have the precise same version of the C++ runtime
 > library as you.
 
 There doesn't need to be a special switch for this.  The way to override
 the choice of libraries is somewhat linker-specific, but something like
 
     g++ .... -lfoo -lbar -Wl,-static -lstdc++ -lquux -Wl,-dynamic -lbaz
 
 will (assuming dynamic libs are the default choice) use libfoo.so, libbar.so,
 and libbaz.so, but libstdc++.a and libquux.a.  The spellings of "-static"
 and "-dynamic" are up to the linker you're using, so "man ld".  On Solaris
 it's something like -Bstatic and -Bdynamic.
 
 
 -- 
 I would therefore like to posit that computing's central challenge, viz. "How
 not to make a mess of it," has /not/ been met.
                                                  - Edsger Dijkstra, 1930-2002