On Thu, Apr 24, 2014 at 3:49 PM, Marvin Renich <[email protected]> wrote:
> I did not see your response right away because I am not subscribed; > please CC me in responses. :-( Not a simple thing, requires laborious copy/paste, not practical on a mobile phone. Maybe you would subscribe to the list? > I thought that because this RFC was still in the draft state that we > didn't need to worry about versioning. I am averse to adding any > meta data to this protocol, as it is a very simple ASCII-encoding > protocol, and making it more complex would reduce its usefulness. We can indeed change drafts, yes. However since RFC 32 has been "deployed to real users", it has to be marked as stable, which I've just done. > My change only suggests handling frames that are not a multiple of 4 > bytes, is backward-compatible, and does not require versioning. If you > would not feel comfortable changing the character set without version > meta data, than I would rather keep the current character set. That would be simpler. The reality is that we'll have to support existing Z85 v1 encoders for the foreseeable future. There is perhaps an alternative to versioning, which is to fix the character set but in a forwards compatible way. That is, a v2 codec will encode using v2 characters, but can decode using v1 characters as well. We cannot reuse an existing character in a different position; however we can deprecate certain characters and replace them with others. > I'll make a pull request, probably next week. It will only deal with my > proposed change, not any change to the character set. The best is to fork the RFC, make a new one, and a pull request for that. I may add some changes to the character set as well. -Pieter _______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
