On 23.03, Laurent Bercot wrote: > On 23/03/2015 14:54, Patrick McHardy wrote: > >What does work is using > > > >-Wl,--whole-archive libnftnl.a -Wl,--no-whole-archive > > > >Not sure if we want to take this path, but it seems acceptable to > >me. > > I'm sorry, but that is not acceptable to me : it defeats > the purpose of static archives, and forces nft, and every > subsequent program using libnftnl, to include in its binary > every possible optional feature you may want to add to libnftnl. > > As a low-level networking utility, nft has its place on > embedded devices (which is my current case). The statically > linked binary for x86_64 (with musl) is already more than 600 kB, > stripped: that is huge. It's more than my whole busybox binary. > And that is with XML and JSON support entirely disabled. I would be surprised if the alternative provides a much smaller binary. Basically everything is referenced through the registered structures and functions called from there. I don't think it makes any difference. > Risking further growth in an uncontrollable way just to use a > gcc-specific feature - which also prevents compiling it with > clang/llvm - doesn't sound like the right solution. > > As a user, Pablo's patch looks like a good fix to me. Well, we would like to keep the callbacks. We can continue to look for a different solution. But either way you will have tons of potentially unused functions in there since the library resolves a lot of things at runtime. -- To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
