On Sun, 2018-06-17 at 19:55 +0200, Niels Möller wrote: > Nikos Mavrogiannopoulos <[email protected]> writes: > > > I attach the current state. It does move all internal symbols into > > multiple internal headers > > Makes sense to me. > > > The last > > patch renames _nettle_md5_compress and _nettle_sha1_compress and > > includes it into the exported list > > I think this makes sense too. We should leave > > #define _nettle_md5_compress nettle_md5_compress > #define _nettle_sha1_compress nettle_sha1_compress > > in the header files, for API (source level) backwards compatibility.
Definitions are already in place. > > If we do that too, can we use glob patterns in the linker scripts, > instead of listing all symbols explicitly? It might make sense to do > this in three steps: I'd suggest against doing that because there will be no way to achieve symbol versioning. See the previous discussion at: https://lists.lysator.liu.se/pipermail/nettle-bugs/2018/007287.html > 1. Rename the above compress functions. > > 2. Move declarations if internal functions to uninstalled headers. These are already done. regards, Nikos _______________________________________________ nettle-bugs mailing list [email protected] http://lists.lysator.liu.se/mailman/listinfo/nettle-bugs
