mouss wrote:
but if you distribute a system that will be used by commercial vendors, things get different. and they are complex because it's not about your interpretation of the license, but about possible interpretations by the vendors/resellers/customers.
Which is, interestingly enough, a good reason _not_ to adopt BSD-style licensing. Doing so would enable GNU to simply fork the entire Postfix codebase and re-release it under a GPL license. If that occurred the FSF would be able to file a lawsuit against anyone who writes but does not publish their own code against the GPL diffs, or at least anyone with deep pockets. Note(1) IBM has deep pockets. Note(2) the FSF has successfully sued Cisco for using GPL code. Note(3) somewhere around half of the Linux code-base is re-licensed from original BSD sources. Note(4) Linux distributions contribute back essentially nothing to the BSD codebase.
the OpenBSD guys take this a bit too "aggressively". on the other hand, this approach has resulted in good software (the so-called OpenBSD pf is a good example, although the story was "special").
This is the interesting part. For all of Theo de Raadt's complaints about GNU "stealing" BSD code (search on "OpenBSD: Stealing vs. Sharing Code") it apparently does not concern him enough to change OpenBSD's license to something more like Postfix, Apache, MIT, or SCSL licensing. Bottom line is that a healthy software ecosystem requires both types of licensing. Personally, I'll be happy if OpenSMTPD just does better than OpenNTPD did. Roger Marquis
