I believe the whole point of the nn_symbol function was to allow for binding authors working with dynamic languages to seamlessly support nanomsg without requiring access to header files which could change between releases. Secondly, the library is still alpha. There should be no concern at all for backwards compatibility at this point. cr On Aug 27, 2013, at 10:11 AM, Martin Sustrik <sustrik@xxxxxxxxxx> wrote: > Hi all, > > As pointed out in issue #108 there are several error codes reported by > nn_symbol() function that are never returned from the library. Namely: > > ECONNREFUSED > EINPROGRESS > ENETDOWN > ENOBUFS > ENOTSOCK > EPROTO > > It would be cleaner to remove these symbols from nn_symbol(). > > However, if anyone things that would a threat to backward compatibility, > please do say so. We'll leave the symbols in place in such case. > > Thanks! > Martin >