On Wed, Mar 2, 2011 at 10:25 AM, Matthew A. Miller <[email protected]> wrote: >> When I release permissive open source code, I save all of you time. >> Even in your commercial projects. >> Also, that's not 8 hour days -- because I have a full time job and I >> only work on XMPP-RTT during my free time. >> Also, it took me less than 1 day to do a test rtt without real time >> editing. 2 days is with editing capabilities. >> > > 1) Unless the code released for a lot of different platforms and languages, > it will not save everyone time; in fact, it will save very few people time, > and may actually *cost* more time if the protocol is under specified. > 2) Unless the license is "public domain" or something as permissive (e.g. > 2-clause BSD), this code will not save commercial projects time. > 3) Some of us like our software to go through rigorous testing, for which "2 > part time days" quickly turns into (at least) 1 work week. > 4) One implementation is not enough for ratification of a protocol. > 5) Relying on an implementation to completely document a protocol is not > acceptable. > > - m&m
Ok, those comments are fair. On the principle of perspective I agree with all 5. That said, I am sure that you agree that source code is better than none at all. :-) Except for the external library (currently using jabber-net and Google GData for the moment), I should note that all the code in the software is either already public domain. Therefore, I do have complete freedom to decide on which open source license to release my client under. Currently, I am considering Apache 2.0. Understandably, the libraries have their own separate licenses. Over time more options will appear, perhaps even by me. But it's a start. Meanwhile, let's observe that the spec is much more complex than its own implementation. I'm now working on that.
