Even though Theo de Raadt is often abrasive, you have to give him credit for making "do it right" cool enough to attract others to his security-first perspective. OpenBSD, OpenSSH (used now in most Linux distros), OpenBGP (complex routing), PF, etc. are all pretty excellent pieces of software.
Yes, I think it's worked out pretty well this way.
Would OpenSSH, pf, git, xorg, most "djbware", Samba, the *BSDs, or the entire GNU project exist if we didn't first have badly licensed software, badly designed software, real or imagined personal slights on mailing lists, etc.?
But instead of "OpenSSH, pf, git, xorg, most "djbware", Samba, the *BSDs, or the entire GNU project" we would have well licensed and well designed software, etc, while something else would have been aweful and people would spend time solving that problem...
The systems that GNU resembles were poorly licensed. How do you figure that without GNU, we would have had a better replacement in their stead (what precedents do you have for your view?)
Oh, I just don't think that we can say that the bad software has been a good thing. Yes, they've made some creative minds produce some really good software, but I still think it would have been better if the first software was awesome right from the start.
After all, the creative minds would still search for unsolved problems…
SMTP strikes me as one of those things where complying with the spec is probably relatively trivial, but accommodating existing, broken software requires significant investment.
http://www.proweb.co.uk/~matt/awk/smtpd.awk is one I did in a couple of hours, and although it dealt with 1,000s of emails in its time, I wouldn't put a price ticket on it!
I wondered that too. Certainly seems like fixing qmail to make it a little less rabidly user-unfriendly would be better than completely reinventing the wheel.
But kudos to them for going out and doing it anyway. I don't know if the world really needed one more MTA but it certainly won't hurt anyone.
For OpenBSD in particular, there's a history of arguments between djb and Theo about the licensing for qmail and djbdns. They're both incredibly stubborn, and while this means they have very high standards, it also means they can be hard to collaborate with.
I was thinking the same thing. I also wonder why more Linux distros do not include qmail. Other than having a few anachronisms, which can quickly be brought up to modern times, it's a fast, secure, time-tested, and completely respectable mail system.
Given that Postfix and Exim have much friendlier upstream developers (qmail barely has a maintainer), I don't think it's a surprise that qmail is losing popularity.
qmail itself is completely unmaintained and patches to it are actually rejected. net-qmail and others have sprung up to fill this void, but its got nowhere near the community surrounding it that Postfix does. Plus, qmail is very hard to use for someone just looking to send some mail, which is the common case (not the case qmail was designed for, either).
Someone should also mention that 'public domain' != free to use. Many countries don't recognize the public domain which means that many projects wont touch code in it. They would much rather have BSD or GPL code.
What I'm trying to say is that public domain has issues within the international realm. With something like the BSD license you declare very explicitly what you allow and I believe (if I understand correctly, I am not a lawyer) according to the Berne Convention that those explicit declarations work in an international realm.
So if your product is never going to be used/sold/shipped out side of USA, England, and a few other countries Public Domain is alright. Otherwise you need to use something like Creative Commons Zero License or a BSD/MIT License.
The Berne Convention is much too old to specifically address the bare licenses used for open-source software. However one of the provisions of the convention is that any creative work is protected under copyright law without requiring explicit registration. This creates a problem for declaring something in the public domain since there is not really an established legal procedure for a copyright owner to waive or forfeit the exclusive rights which have been automatically granted to them.
postfix was designed by one (that is very important) talented engineer - there is an excellent well-balanced architecture behind it along with professional implementation. It is much better to ask the author to make a special release with different licence and then import it.
It is much better to reuse an excellent code of great projects like nginx, postfix or dovecot than [re]write it from scratch.
