On Fri, Jun 13, 2003 at 04:52:39PM +0200, Robert Millan wrote: > On Fri, Jun 13, 2003 at 08:01:47AM -0600, Joel Baker wrote: > > On Fri, Jun 13, 2003 at 02:05:12PM +0200, Robert Millan wrote: > > > On Thu, Jun 12, 2003 at 10:17:14PM -0600, Joel Baker wrote: > > > > > > > > And the NetBSD one is *not* GNU-based, > > > > > > then it doesn't make sense to call it "GNU/NetBSD", would you mind > > > fixing the website accordingly? > > > > It still uses a primarily GNU userland > > You just said it is "*not* GNU-based". Do you know what GNU/Something means? *sigh* It was early in the morning. I typed GNU on autopilot; I *meant* "GNU libc & other stuff". As for knowing what GNU/Something means - does *anyone*? Nobody could explain the details when it was first debated, and to date, nobody has explained them here, either. Or, to put it mildly, I suspect any attempt at doing so may well be met with "I do not think that means what you think it means", just because there does not appear to, in fact, be a definition of it. > > - coreutils is there, just like > > almost every other Debian package. There was some debate over the precise > > name, when it first came up; however, as far as we can tell, the request by > > RMS that Debian agreed to involved the userland tools, not just the libc. > > RMS would never request placing "GNU/" in the name of a system that is > "*not* GNU-based". Maybe. Maybe not. He requested it of systems that didn't consider themselves GNU based, but *he* considered them to be. Debian agreed with him, others didn't. C'est la vie. > We had this debate over the NetBSD-based port before but it obviously > didn't bring any productive results. I don't claim that it is or it isn't > a GNU variant, but at least the port maintainers should have a clear opinion. The port maintainers have done their best to try to ascertain what, exactly, the GNU/* means, and have failed to get useful answers, yes. So. I present to you the following system; you tell me if it's GNU, or not, and if at all possible please cite references to RMS (or at least FSF folks) asserting any particular point. 1) Non-GNU kernel (IE, anything but Hurd) 2) Non-GNU libc (anything but glibc) 3) A userland whose core utilities (ls, rm, cat, etc) come from GNU coreutils, fileutils, findutils, and similar packages. 4) A userland whose primary compiler is GCC, and whose primary make is GNU Make, but which uses another make variant for buildin kernel & libc. 5) The rest of the userland may be under any DFSG license, and may or may not have anything to do with the FSF or GNU stuff. That's the situation with the NetBSD port as it stands. I'm happy to discuss whether it should be "Debian GNU/NetBSD/i386" or simply "Debian NetBSD/i386", but if we're going to dredge this up again, I'm going to have to insist on someone providing actual citations, rather than assertions (well, okay, RMS or FSF folks could assert it, since that's what the citations would be to...) Keep in mind that the rest of Debian might also have a say, though I strongly suspect that as long as we have clear reasoning, nobody will care much which way it goes. -- Joel Baker <fenton@debian.org>
Attachment:
pgpxL16MUELB6.pgp
Description: PGP signature