« GPL v3 Released | Main | Start the stopwatch »
20060117 Tuesday January 17, 2006

Spreading GPL across SOA and AJAX?

License Giants

Just back to my hotel after the second (final) day of the GPL v3 conference. I attended the Q & A panel (pictured to the right) in the morning, discussion committee B around lunch time (representing Sun) and discussion group A this afternoon (representing OpenSolaris). Today has been much more about getting doen to the gritty work of adding comments and discussion to the draft. Indeed, group A added a number of comments this afternoon and the comment flow is roaring ahead (fortunately there's a syndication feed).

I notice that Stephen O'Grady at Redmonk thinks like Mike Olson that the new GPL funked out on the provision to rein in ASPs. At first I thought so too, when I saw that there was no statement extending the need for publication of source for web execution of GPLed code and that it seemed to be a goal of section 7:

Aside from additional permissions, your terms may add limited kinds of additional requirements on your added parts

I think it's clear that the intent is indeed to initiate a marketplace for extensions. But as I have looked at it more and more, I believe Eben and Richard have been far more subtle. A crude and explicit ASP clawback would have raised a riot. Instead, the seeds are sowed in section 5c:

If the modified work has interactive user interfaces, each must include a convenient feature that displays an appropriate copyright notice, and tells the user that there is no warranty for the program (or that you provide a warranty), that users may redistribute the modified work under these conditions, and how to view a copy of this License together with the central list (if any) of other terms in accord with section 7.

and even more in section 1, "Complete Corresponding Source Code":

For example, this includes any scripts used to control those activities, and any shared libraries and dynamically linked subprograms that the work is designed to require, such as by intimate data communication or control flow between those subprograms and other parts of the work, and interface definition files associated with the program source files.

Doesn't that latter language just yell "SOA" to you? It seems to me that there is the germ of an idea here to ensure that, if your server software is licensed under GPL v3, the client software - AJAX, plug-in or whatever - is required by this language to also use GPL v3. This is a much more subtle approach, and strikes right to the heart of the rapidly evolving "read-write web".

By the way, if you're looking for a properly formatted version of the GPL v3 draft, Tim has created one. [Update: He's added paragraph-level links so I have linked sections above]

Posted by webmink ( Jan 17 2006, 06:41:43 PM PST ) Permalink Comments [3]
technorati del.icio.us digg slashdot
Comments:

Interesting analysis, Simon. I disagree; it seems to me that there are straightforward ways to close the ASP loophole, if that were what the FSF wanted to do. More fundamentally, I don't believe that any back-door mechanism can possibly survive the public comment period. I've posted a follow-up in my blog.

Posted by Mike Olson on January 18, 2006 at 04:02 AM PST #

I *thought* I got an answer to this question when I asked Eben whether software delivered through a managed hosting service counted as distribution or propagation. (This was the afternoon of the 1st day--I don't think you were there at the time.) He gave a fairly long answer but I understood him to say that, so long as software was being delivered as a shared service (e.g. not dedicated servers for each user), that was not distribution under copyright law. But perhaps he answered one question about SaaS, but not all of them. I agree that if there's any ambiguity, it needs to be cleaned up.

Posted by Gordon Haff on January 18, 2006 at 06:10 AM PST #

Gordon: I think this is a different issue to "propagation", Gordon - it relates rather to the definition of "the Complete Corresponding Source Code for a work". I agree that the definition of propagation does not extend to ASP usage, and that's what I was referring to as the crude approach that was not taken. However, I am of the view that the section 1 language at the very least must open for discussion whether SOA usage is a vector. As Mike suggests, this will definitely be discussed during the public review.

Posted by Simon Phipps on January 18, 2006 at 07:01 PM PST #

Post a Comment:

Comments are closed for this entry.