;D
I’ve tried to make clear that I know full well I’m only going on personal opinion and limited experience on this particular point.
Whereever it sounds like I was saying “I’ve looked at everything on SF” what I meant was “Much of my exposure to os projects is via SF (as opposed to via freshmeat, PBONE, etc - just so you know what I’m using for source material; if you think sf is a biased source, you can say so directly and hopefully point me somewhere where I could get a better view), and it’s quite a good place to look when trying to get an overview of opensource projects”.
Over the years, I’ve tracked many SF projects. I’ve downloaded and tried something like 300 of them from SF, max. (Out of approx 100k currently registered at the moment, I think? and they haven’t been deleting any for most of that time…). I’ve downloaded and used probably a similar number from other sources (I’m not including anything I’ve got from distro CD’s here - only stuff I found or went looking for). I certainly haven’t looked at all the source for all 300 sf projects I’ve used, but I’ve looked at snippets of many, and looked in detail at quite a few, and even looked at the entire source of all files in detail for a dozen or so.
I can only say that I’ve seen what I identify as a pattern (well, several patterns), in terms of the quality of software engineering in os software. Incidentally, another interesting one I’ve noticed is that it’s only when the major / standard linux app or driver in one area completely stagnates for about 4-5 years that someone comes in and makes a much better one; if the current app or driver is “bad but even slighlty active” (even if it doesn’t fully work for many people!), that seems to be enough to prevent others competing. Off the top of my head I’d say the opposite was true in the commercial world - poor software breeds competitors wanting to take away their money like nothing else. In os, most examples I can think of of rampant competition is where each of the competitors is about as good as each other, but duplicates most of each other (KDE and Gnome; XMMS, Xine and Mplayer; etc etc).
You can also get a really good flavour of the quality of software from looking at the planning and documentation (the link between good docs and good app is not trivial, but I hope everyone understands what I mean, whereas the link between good planning and good app is much more obvious). I’ve looked at the planning and docs for many many projects. I’ve complained about the docs, and offered suggestions of particular aspects that need additional docs for quite a lot of OS projects. I’ve even been responsible (indirectly) for updating of a few HOWTO’s, including one of the kernel ones.
None of this qualifies me to be correct, and I’m only mentioning it now because it’s what I was thinking when I made the comments you (fair enough) called me out upon; I attempted to allude to it, rather than listing my precise experiences and analyses (I’ve also done the odd report on os software for people to make support or purchasing decisions) of os software.