On 16 Sep 2014 11:32, "Daniel Stenberg" <[email protected]> wrote: > > On Mon, 15 Sep 2014, Peter Stuge wrote: > >> You make OpenSSL calls and only add the functionality to the openssl.c file so your patch can not work with anything but OpenSSL, and I for one want that addressed before I consider the patch ready for inclusion. > > > I have a slightly less rigid attitude: as long as there's a good story about the alternative backends, that they don't break the build and the limitation is documented, then I'm fine with an implementation that "heads further" with one of the backends than the other(s). If not, we'll get a situation where the multitude of backends will hamper innovation as users will just not do things.
I agree. In fact, we're already in this situation. For example, complied against OpenSSL the user can pass just the private key file, but with libgcrypt they must also pass the public key. Alex
_______________________________________________ libssh2-devel http://cool.haxx.se/cgi-bin/mailman/listinfo/libssh2-devel
