On Fri, 2018-06-08 at 10:41 +0200, Niels Möller wrote: > Nikos Mavrogiannopoulos <[email protected]> writes: > > > Niels, I'm not sure however if that was your intention. Didn't you > > want to deprecate some of the _nettle symbols as well like > > _nettle_secp_256r1? > > I was thinking of doing something simple, with nettle_* symbols going > into the supported ABI (symbol version NETTLE_@LIBNETTLE_MAJOR@), and > all _nettle_* symbols getting symbol version > NETTLE_@LIBNETTLE_MAJOR@_@LIBNETTLE_MINOR@, which explicitly makes > them > *not* binary compatible between minor versions.
> I think it's helpful that ABI status corresponds to the names used in > the source and header files, both for maintenance and for user > documentation. If you mean removing them from the public headers and placing them in one (or multiple) internal ones, it makes sense to me. Otherwise it would be confusing on why they are listed in headers if they are unavailable or break ABI. > What do you think? Are there any of the current _nettle_* symbols > that should be in the advertised API (and hence renamed)? I do not use any of them in gnutls, but searching at the debian code, I see: _nettle_md5_compress (sogo), _nettle_sha1_compress (filezilla/putty) regards, Nikos _______________________________________________ nettle-bugs mailing list [email protected] http://lists.lysator.liu.se/mailman/listinfo/nettle-bugs
