Posted by Sacha in English, IT.
I’ve never been at ease with GPL-dual licensing business models. What worried me wasn’t so much that OEMs had to pay for a proprietary license on the product – OEMs spend half of their lives licensing software and the other half paying royalties, FOSS or not. No, what I didn’t really like was the viability of this model, or more precisely that it invalidated one of FOSS’s key value and promise: the extreme viability of the code. Let me explain…
The good thing with FOSS-backed business models is that as a user of a FOSS product, whatever happens to the company behind the product (getting evil, going bankruptcy, etc.), you do have a B-plan. That B-plan might not be ideal, it might take time to materialize, might cost some money, but there *is* always a B-plan.
The problem with GPL dual-licensing business models is that this is simply not true. If the company behind a dual-licensed product X get sold to an evil company, goes bankrupt or decides to increase its prices by 400%, all OEMs are doomed. In the best case (i.e. if the IP is still owned by some legal entity, read: no bankruptcy), the OEMs can keep paying a high fee. But in case the IP owner just go bust, you cannot license that IP from anywhere anymore and you are only bound by the terms of the GPL and either have to GPL your own product or get rid of that viral piece of code. Don’t get me wrong, that situation is very similar to any proprietary product, but it then defeats one of the key added value of FOSS software: by going the GPL-dual-license route, OEM customers do not simply pay a fee, more importantly they loose one of the most important feature of FOSS: long term viability of the code.
So, to go back to the MySQL/ORCL drama, all of this bla bla wouldn’t take place if MySQL wasn’t dual-licensed (but was LGPL based for example): all customers (OEMs or end-users) could keep using it as-is, with no restrictions as to whether they have to open source their own product or not. Granted, they would loose the support/maintenance provided by SUN/MySQL, but could hope to find quality support from another company. There would be no ORCL/EU story, nada. For example, this problem doesn’t exist for other SUN-owned FOSS products such as Glassfish (CDDL) or OpenJDK SE (GPL+CP-Exception).
Oh, and if ORCL wanted to be (really) evil, they could simply relinquish control of the project governance to some “open body”, give away the trademark of the project and walk away… but KEEP MySQL’s copyright in their safe. They would then argue that they are good FOSS citizens since they have relinquished control on anything FOSS-related. This would obviously kill MySQL’s OEM market but ORCL could hardly be criticized by Open Source proponents for that move: the problem would lie in the absence of a *proprietary license* here, not in any FOSS-related concern. And if needed, this could be properly “spinned” as FUD by ORCL PR machine (and MSFT for that matter) to show how weak the Open Source model is – when it actually only shows *the* big weakness of dual-licensing models.
Some of the ex-MySQL co-founders who now ask for ORCL to let MySQL go are responsible for the current situation: their choice of a dual license business model years ago is what led to the current situation … but also what led MySQL to a 1B valuation. You cannot have it both ways I guess.
For a free marketing ride on the FOSS buzz, the dual-licensing model could cast a FUD-cloud on all FOSS-based business models.