Re: Favoring static linking at configure time

[Peter Rosin]

On Sun, Dec 16, 2007 at 04:23:41PM +0100, Benoit Sigoure wrote:
> On Dec 4, 2007, at 9:55 PM, Ralf Wildenhues wrote:
> 
> >Hello,
> >
> >* Brian Dessent wrote on Tue, Dec 04, 2007 at 07:20:40PM CET:
> >>Benoit Sigoure wrote:
> >>
> >>>>Of course there is: -Wl,-Bstatic -lfoo -Wl,-Bdynamic
> >>>
> >>>Hmm I didn't know this.  How portable is it?  I guess a gccism, but
> >>>is it even portable across versions/ports of GCC?
> >[...]
> >>-B seems to be fairly portable; all of the following support it:
> >
> >I implemented support for it in Libtool a while ago:
> ><http://thread.gmane.org/gmane.comp.gnu.libtool.patches/6727/ 
> >focus=6730>
> >
> >it was postponed due to "2.2 being out real soon now", and it being
> >possibly destabilizing to HEAD.
> >
> >Maybe some data points to the end that it's rather stable (I think I
> >tested it pretty well at some point) would help.
> 
> 
> OK So finally I solved the whole thing by adding /abs/path/to/libfoo. 
> $libext (instead of -lfoo) in $LIBS ($libext is computed by libtool,  
> most of the time it's .a but it happens to be .lib at least for  
> MSVC).  This is probably fragile but at least it works on GNU/Linux  
> (GCC, ICC), OSX (Apple GCC) and Windows (Cygwin, MinGW, MSVC).

It is fragile. There are several ways to use MSVC with libtool. The
patches I have worked on does not name the static library according
to your pattern, so your scheme would break with them.

I don't know if you care though...

Cheers,
Peter