Hi Tyler
More comments inline:
On Sun, Feb 8, 2015 at 6:11 PM, Tyler Romeo tylerromeo@gmail.com wrote:
Not sure exactly what you mean. The MPL is compatible with both GPL 2.0 and 3.0.
MPL v1.1 was not compatible with any version of the GPL, hence the reason why Mozilla eventually multi-licensed their software under MPL/GPL/LGPL (they pioneered the approach, IIRC). MPL v2 achieves GPL compatibility by specifically naming GPL v2+, LGPL v2.1+, and AGPL v3+ as "Secondary Licenses", and then optionally allowing relicensing (i.e. one-way conversion of the software to GPL/LGPL/AGPL) under section 3.3. It basically bakes multi-licensing into the license itself.
What I'm saying is that GPLv2 to GPLv3 is a one-way direction. Once code is licensed under v3, you cannot go back to v2. The Apache license is only compatible with v3, and code under v2 cannot be combined with Apache code, so the only way to add in Apache code to a GPL project is if the entire project is v3.
By this logic, once MPL-licensed code is relicensed under GPL, you cannot go back to MPL, so the only way to add GPL code to an MPL project is if the entire project is GPL.
The fact that there's an option to upgrade the license to GPLv3 is sufficient for GPLv2+ code to be compatible with Apache 2.0 code, in the same way that the option to relicense MPL code as GPL code is sufficient for GPL compatibility. You don't need to actually perform the relicensing, you merely have to offer the licensee the option to do so if they so desire.
I will assume good faith for the WMF. I was just making a quick jab; I know the WMF is not going to make MediaWiki proprietary.
Thanks for clarifying that.
However, I will not assume good faith for every other software company out there that may take MediaWiki, modify it or improve it in some way, and then begin selling it as proprietary software. It's nice to think the world is an ideal place where everybody shares their source code, but unfortunately we are not living in the ideal, and in fact that is the entire reason the GPL was written in the first place: in response to companies acting in bad faith.
I think my experience is more-or-less in line with Stas, though I think I still have more sympathy for the copyleft approach than he has. I see the copyleft/permissive choice as more of a tactical than a moral choice, and that I think permissive is going to be a more effective long-term tactic to achieve the moral goals also sought by proponents of copyleft. Downstream modifiers of software eventually learn that keeping secret patches to upstream code is more hassle than whatever "secret sauce" benefits they get.
Also, one cost of copyleft licenses is that they are much, much more complicated than permissive licenses. Even though many people feel comfortable with the compliance requirements of most OSI-approved licenses, the permissive licenses can usually stand alone without an FAQ, whereas an FAQ is required for just about all of the copyleft licenses. That simplicity reduces a very real barrier to adoption.
Rob