For instance, Linux was used in a Linksys router, which forced them to release some code (though mostly blobs) related to the Broadcom drivers.
It was a binary driver, they were not forced to release the source code to it, only the source to the rest of the FOSS software they were using.
The more innovative your gadget, the more likely you'll have to write a driver, and the more likely you might be to want to keep that driver secret
Which you can do with Linux.
private forks are irrelevant to this discussion since they don't need a license anyway.
They are very relevant, because they are a "loophole" in the GPL
Not being applicable to internal usage was a deliberate intent of the GPL, it is only concerned with distribution to the larger public.
But if you insist, consider nVidia. They seem to have found a loophole which lets them insert their own code into the kernel.
??? That is not a 'private fork'. You seem to have changed subjects again...
The only difference here is that they are given a choice between the risks (and rewards) of contributing, and the risks (and rewards) of maintaining a separate branch.
Thats obvious. My only point is that for some companies the risk/reward of a GPLed OS makes better sense for them.
Basically, what you're claiming, hypothetically, is that something completely different -- not Linux, not HURD, and not BSD -- would've been invented,
Not completely different in the technical sense, only in its licensing.
and would've surpassed BSD, which would clearly dominate, at least for a short time.
Note that I never said 'surpass' or 'dominate'. Only that those for whom the licensing works better would switch to it eventually. Just as those for whom the BSD works better already have been, or will be in the future, switching to the BSD over time.
And that is where we differ, because I really don't see that happening.
Fair enough. That part of my post was a 'hypothetical of a hypothetical' for which I have no evidence.
for example, the reason projects like Postfix are successful is because
You keep going back to non-OS examples, when my point was all about licensing issues related to operating systems...
And you keep ignoring my point that neither the GPL nor the BSD work for everyone, so obviously there will be plenty of examples for each side...
KHTML probably never would've made it anywhere close to Gecko
Becoming Webkit was never its goal, it was just KDE's internal HTML renderer, which was then taken by *others* and turned into something more than even its original designers had intended.
So you're right, this is where we do differ -- because BSD does work better for other companies,
Sigh... when did I ever say the GPL worked for everyone? All I said was the corollary of your above point is also true. Thats why the GPL and BSD will always both be around.. because neither one works best for everyone.
What I'm arguing is that there are plenty of people who choose Linux today, even though it's not the license they'd prefer,
If there are enough of those people to form a large enough dev community for the BSDed OSes, then the BSDs will eventually catch up to Linux, and those people will switch.
My point is only that you shouldn't expect Linux to die at the point, since, for some/many people, its licensing works better for them (again, a corollary)...
I realize you're claiming it wouldn't get that far.
Not at all, you keep making the mistake of assuming I'm one of those 'GNU zealots' who hates the BSD and wants to see it die.
My only point from the beginning was that the 'BSD zealots' who believe that Linux and the GPL will die once they've caught up to it, are simply deluding themselves. There is a reason both exist and are being actively used.
don't underestimate how significant that headstart can be. From this article:
And yet C has been eclipsed by many other languages since then, in terms of 'popularity'. On the other hand, its still around and being used because its still useful to some people. Really, I don't understand your point here.
What you're arguing is basically that the BSD license is so wrong for kernels
No, that's what you think I'm arguing. You're not reading what I'm actually saying.
I think I'm done here.