Proprietary Feedback Part 2
Two weeks ago I explained how proprietary software might benefit open source projects, and began to look for examples to support a proprietary feedback hypothesis. As Paul Ramsey wrote, it's hard to get proprietary users to come out of the closet, so it's very likely that I don't have all the evidence. What I have found does not support the hypothesis of a general feedback between proprietary and open source GIS software. Any feedback is much more narrow and selective. I now have 2 new hypotheses:
Proprietary benefit for open source GIS software is primarily a phenomenon of the GDAL project. No other project has such a list of proprietary contributions, and I think this says more about Frank Warmerdam, his ingenuity, and his customer service than it does about open source GIS in general.
Proprietary benefit for open source GIS software goes almost exclusively to low-level projects. What are proprietary software companies contributing to GDAL? Drivers. Format readers and writers. Bedrock functionality. Not all contributions go to GDAL (a fact that undermines my first hypothesis a bit), but the others do go to similarly low-level projects. See the comment about GEOS in my previous post.
My conclusion is that developers of higher-level open source GIS projects shouldn't be overly concerned about scaring off contributions by choosing the GPL. I've found vanishingly little evidence that proprietary companies contribute at other than the lowest level.
Comments
Re: Proprietary Feedback Part 2
Author: Chris 'Xenon' Hanson
I would agree with the facts for the most part, but I think the conclusion is subtly flawed by the choice of the investigation. High-level apps are less likely to contain code that is useful in a context outside of themselves (open source, or not). One might as well ask, what portion of code from a high-level open source app is reused in ANY other context -- open source or not. It's simply a matter of common sense that low-level libraries and components are more likely to be useful to projects outside of themselves, open source or proprietary. The greater outside use a body of code has (irrespective of the licensing it's used under) the more likely it is to get contributions from those users (again, completely regardless of what licensing those users participate under). If you compared the amount of code reuse between a high-level apps and low-level apps, I think you'd find it's in remarkably similar in proportion to the same comparison focusing only on proprietary users/contributors. A side issue is this: It benefits users, commercial and otherwise, to have commercial/proprietary companies utilizing and contributing to low-level Open Source projects. The more people who utilize a piece of code (especially something implementing a standard like a file format) the better exposure and expansion of capability that code gets. As a concrete example, look at libtiff/geotiff or libjpeg. So many programs rely on them that it's to the point where if those tools can't handle a particular file, it's probably a failing of the file. The open source nature of the code has led to them becoming the de-facto standard, improving interoperability far beyond what we'd have if everyone tried to roll their own. A discussion of the pitfalls of a code monoculture (a la Microsoft Windows) is beyond the scope of this discussion.Re: Proprietary Feedback Part 2
Author: matt wilkie
What about ghostscript? It's the oldest open source project I can think of with a very strong proprietary backing. Dunno how many proprietary contributors there are outside of Aladin tho. Although it's model is very different from most, and from gdal, it's history also supports the low-level hypothesis.