Proliferation of Proliferations
By webmink on Feb 18, 2005
Seems open source license proliferation is the topic of the week - obviously some memo went round that I didn't get. Here are the paths I have seen proposed so far to end license proliferation:
- Just use the GPL for everything (and kill the rest). This is actually not a solution as I explained in my earlier essay. The GPL is not miscible with other licenses and thus offers no path for open sourcing software of mixed heritage. In addition, it does not have a strong mechanism to deal with the issue of protecting developers from patent indemnification issues. Both of these will get fixed I hope when licenses "proliferate" a little more with the creation of GPL v3.
- Analyse the licenses and build a chart showing which licenses have which effects so that new licenses appear less necessary and old ones can be recycled. SFLC are offering to do this and it will prove useful to lawyers seeking to license new projects, but the effect it will have on proliferation is small if there's no license model that actually meets their needs.
- Close down OSI, or at least prevent it approving any more licenses. The latter seems to be what Martin Fink from HP and SJVN want to do, but it would solve nothing. Licenses proliferate because existing ones don't fit people's needs (or at least so their lawyers tell them), not because OSI exists and keeps approving new ones (or as a rapper might put it, "licenses don't kill software, lawyers do"). Having said that, the new ideas OSI is having for including necessity as one of the approval criteria are good.
- Create a small set of generic, miscible licenses that need no modification to suit most needs and which can be gently modified without creating new non-miscible 'fragments' of the open source code space. This is what CDDL set out to do, and I believe what GPL v3 sets out to do, and is in my view the best solution to license fragmentation. It's also something that OSI could adapt itself to regulate.
So why are people (and there are plenty) who know better trying to paint CDDL as part of the 'license proliferation' problem? Are they too afraid to be positive about it just because Sun had it written (largely by non-Sun open source experts, by the way). If the term 'proliferation' applies to CDDL, it will also have to be applied to GPL v3.
CDDL was designed from day one to be a possible solution to the problem of license proliferation, which is largely the result of corporate lawyers wanting to use the MPL for their software but being unable to do so because of issues with the license. CDDL intentionally deals with those issues in a generic way, so that others can use CDDL where they need an MPL-style license. Indeed, John Cowan pointed to the license during the approval discussion at OSI as a replacement for the MPL and went on to justify it.
So it can't be that the CDDL is part of the problem - it's way too much part of the solution for that, as John Cowan just asserted. Maybe the people slinging mud at CDDL are actually slinging mud at Sun, for some other reason? Maybe there's another agenda? The politics of competition, maybe?