Static linking considered harmful
Jakub Jelinek
jakub at redhat.com
Wed Nov 22 11:08:56 UTC 2006
On Wed, Nov 22, 2006 at 11:57:11AM +0100, Olivier Galibert wrote:
> On Wed, Nov 22, 2006 at 11:49:04AM +0100, Patrice Dumas wrote:
> > That's not true for numerical models. I don't really underestand what you
> > are talking about. When I compile a model with
> >
> > gcc -static mymodel.o -lnumerical-lib
> >
> > I don't care at all about system share libraries, this is not a shared
> > executable, so no trouble with new system shared libraries. And once more
> > for numerical models iconv, locales, and nss are not really usefull.
>
> Actually, given the incompetence of the glibc developpers, you do.
> They use shared objects whether you want them or not. You can't
> resolve localhost for instance, which is quite annoying for X
> applications.
And how do you want to resolve localhost when on one box you use
files for that, on another NIS, on another NIS+, on another LDAP,
on another Berkeley DB, etc.?
Even if you only wanted to put files parsing into statically linked
programs and do the rest dynamically, what would e.g. with
#215283, where FC6 anaconda now writes localhost entry as
::1 localhost.localdomain localhost
instead of 127.0.0.1 and gethostbyname would just fail in that case?
When libnss_files.so.2 is used, this is fixable (and will be fixed
soon), otherwise you'd have to recompile all your statically linked
binaries?
Jakub
More information about the devel
mailing list