Re: mklibs and brltty
On Tue, Sep 23, 2003 at 03:31:20PM +0200, Mario Lang wrote:
> Bastian Blank <firstname.lastname@example.org> writes:
> > On Sun, Sep 07, 2003 at 09:04:39PM +0200, Petter Reinholdtsen wrote:
> >> Could weak symbols solve the problem? I'm not very familiar with
> >> shared library internals, but I believe a weak symbol will be selected
> >> from the library if no other matching symbol is available. If the
> >> library provides such symbol, it would make mklibs happy, while the
> >> shared library linker would still select the symbol from the
> >> executable. Is this correct? I am mostly guessing based on the
> >> rumors I've heard about weak symbols.
> Hmm, can you give me a hint on how one is supposed to
> define a weak symbol? This is all rather new to me.
> > dropping subdirs which aren't referenced via rpath from the resolver in
> > mklibs. none of that libs will be reduced.
> I am sorry, I do not grasp this sentence in the given context.
> The problem is not reduction, at least not AFAICS. mklibs is not
> supposed to reduce /lib/brltty/lib*.so. It only checks that
> all referenced symbols are actually defined somewhere (in a library),
> which fails since the missing symbol is actually defined in
> the executable that loads the lib.
Would you please update bug 211092 with this information?