Letter to the Editor: A touch of Insanity at the FOLK's home...Aug 04, 2001, 23:30 (7 Talkback[s])
Jonathan Day writes:
The FOLK project, over on Sourceforge, is collecting a lot of patches to the Linux kernel. Schedulers by the dozen, drivers for a wide variety of devices, support for VME and real-time hardware, more debuggers than you can shake a stick at, a bunch of file-systems and assorted low-latency patches.
The idea of FOLK was always to show-case technologies by developers, in a combined environment, so that side-effects can be detected and that people's imaginations can play with the possibilities, without having to first play with the reject files.
However, there has been one very unexpected side-effect, so far. The most recent FOLK patches exceed a third the size of the entire kernel. That suggests that there is a LOT of unknown/unused/dead code out there, for Linux. And I don't believe I've even found a fraction of what's out there.
Does this mean that there is a problem with how the Linux kernel is developed? No. I don't believe so. The approach followed by Linus Torvalds of including only patches of sufficient quality and which have already received ample testing seems to work jusrt fine.
This begs some questions, though: the patches I've found are no toy gimicks. They undoubtably -have- been tested. So, why are they still floundering around the edges of nowhere? Why aren't they being announced? How can something so GPLed be so invisible?
This, finally, brings me to the insanity of this piece. It is insane to imagine that programmers and software engineers can ever be taught to communicate outside of their niche, or that showcasing alone can ever achieve any real level of integration. As the FOLK maintainer, I can tell you what others have done. But I can't get those others to tell you themselves. That is beyond me.
And this is why I'm writing this. I hope that someone can fill in that missing piece. If communication were adequate, these projects might still be "niche", but they'd be known, tested, maintained, current, and as easy to install as the lm_sensors2 code. As it is, the best I hope for, on finding new work, is one or two out of the five. Most often, it's "none at all". Please, if anyone out there can convince the coders of the world to speak, do so! Cure the insanity at the FOLK's home!