Interesting article, but I think it misses the most important form of optimization of all - data representation. Often the way that you represent the problem in data structures determines the algorithms that are available to you, and thus how fast the problem can be solved. There are a number of examples of this described in books like 'Programming Pearls' and 'The Mythical Man Month'. My favorite is perhaps a bit esoteric, but a good example nonetheless - use of the method of orthogonal collocation to collapse a partial differntial equation to a set of ordinary differential equations. This nifty mathematical trick alows you to reduce the dimensionality of a problem by one, or reduce the big O by one - i.e. speed up the solution by a factor of 10.
Finding a clever data representation is often the lightning bolt that shatters the oak tree.
A bad idea.... VERY bad.
I really don't like this idea, for MANY reasons.
1. How can you even really have free software when the development tools that are needed (i.e. MFC and VC++) are closed?
2. Are you sure that there are no license gotchas associated with VC++/MFC that make the GPL invalid?
3. Are Windows developers really ready for GPL? Or will we be treated to Windows developers incorporating GPL code into their own closed binaries because they thought GPL=public domain?
4. Ports to another system tend to look crappy. This may create the impression that OSS is crappy in the minds of Windows users, and make OSS on Windows a flop. Microsoft may exploit this in it's FUD.
5. I thought we wanted to get Windows developers writing for OSS operating systems, NOT OSS developers writing for closed operating systems!!!! This model could REDUCE the amount of OSS software for OSS systems. Rmember, ATTRACTING DEVELOPERS TO WRITE FOR AN OS IS *THE* KEY TO SUCCESS!!
6. Microsoft is known for anti-choice business practices. Do we really want to support a company that puts out stuff like the Mindcraft Benchmarketing, the Halloween Documents and the Linux FUD page?
7. I don't think we want to soften the financial pain associated with chosing a non-free operating system like MS Windows.
8. This hybrid closed OS + OSS application model gives MS an embrace-extend opportunity into OSS that we DON'T want.
9. Initially the amount of free software will be small compared to shareware, nagware, etc. This small amount may be used as part of a FUD campaign to discredit the OSS development model.
10. Currently free (beer) software on Windows is free because the quality is too low to sell. OSS software could be tarred by the same brush very easily.OSS software is fighting an image of 'a collection of midnight hacks'. The last thing we need is to reinforce that.
11. OSS needs killer apps that are exclusive to OSS operating systems. The last thing we want is 'oh that's available for Windows too, so let's just keep running Windows'.
The more I think about it, the more I think that this is the worst thing that could be done to the OSS community.