From mib@geech.gnu.ai.mit.edu Wed Nov 3 22:09:43 1993 Path: usenet.ee.pdx.edu!cs.uoregon.edu!ogicse!emory!nigel.msen.com!spool.mu.edu!howland.reston.ans.net!usc!bloom-beacon.mit.edu!ai-lab!life.ai.mit.edu!mib From: mib@geech.gnu.ai.mit.edu (Michael I Bushnell) Newsgroups: gnu.misc.discuss Subject: Re: kernel priorities Message-ID: Date: 3 Nov 93 21:29:52 GMT Article-I.D.: geech.MIB.93Nov3162953 References: <199311031439.AA22172@terminus.cs.umb.edu> Distribution: gnu Organization: FOO Lines: 76 NNTP-Posting-Host: geech.gnu.ai.mit.edu In-reply-to: karl@cs.umb.edu's message of 3 Nov 1993 09:40:00 -0500 In article <199311031439.AA22172@terminus.cs.umb.edu> karl@cs.umb.edu (K. Berry) writes: The purpose of the GNU project is to develop and distribute a ``complete Unix-compatible software system'' (quote from the Manifesto). As we all know, the only major uncompleted part of GNU is a kernel. But as we also know, there are two free kernels available -- Linux and 386BSD. Something seems wrong here to me. Whether the FSF should distribute one of these kernels is not clear; we might, we might not. It's a policy decision that (blessedly) I can opt out of. The reason I've heard that Linux was deemed unsuitable for GNU was that it was unportable, mostly because it makes heavy use 386 memory protection instructions. My impression is that this is what Linus told rms, and I don't doubt its truth, as far as it goes. Certainly Linux could be made more portable. It might take work, but it's doable. More on that later... But is this enough of a reason to abandon Linux? I've heard (unquotable rumour) that Linux is being ported to the Alpha, which is surely about as far from a 386 as one can get. More importantly, the entire Linux kernel is (I believe) smaller than the system-dependent parts of the microkernel + the Hurd + the GNU C library. And finally, the Mach micokernel still hasn't been ported (please correct me if I'm wrong) to any interesting new hardware (e.g., Sparc 10's, Alpha, HP 700's), and runs only on x86's (no portability gain there!) and DECstations (ho hum). There are many more free Mach-3 ports than only those two, as it happens. The HP700 port is done or nearly so at the University of Utah. There is an alpha port, free, either done or nearly so. I don't know about Sparc 10's, but I'd expect that to follow soon. I wouldn't take the time to write this message if I thought the Hurd was anywhere near being released. But I don't believe it is (again, please correct me if I'm wrong). And once it is released, like all software it will be unusable for mere mortals for a time. Linux has already been through the shakedown period and is happily being used by many people. My understanding is that the FSF is devoting around two full-time people to work on the Hurd these days. That much effort could go a long ways towards improving and porting Linux. The transparent filesystem jazz and other neat ideas from the Hurd could perhaps be adapted to Linux. The "transparent filesystem jazz" could not be adapted to any monolithic Unixoid kernel; it's just not doable. If the FSF had started writing a kernel at the same time they started waiting for Mach, GNU would probably have been completed and spread like wildfire. These days, there are too many proprietary Unixes for any single one to dominate. A policy decision was made between finishing a free BSD and the Hurd. That decision was made with the knowledge that the free BSD would be finished much sooner. However, GNU's goal is different from a proprietary company's goal: we haven't failed because others have done the free BSD work (both NetBSD and Linux fall into this category). Rather, everyone has won. Secure in the knowledge that other people would do free BSD work, we decided to do something New And Different. That something New And Different is, I believe, sufficiently better than monolithic Unixoid systems that it's worth having out, and will compete *better* than the free monolithic systems can against proprietary systems. Keep in mind: the contest is between free software and non-free software, not between different sorts of free software. -- +1 617 623 3248 (H) | He shall give his angels charge over you, +1 617 253 8568 (W) -+- to keep you in all your ways. 1105 Broadway | They shall bear you in their hards, Somerville, MA 02144 | lest you dash your foot against a stone.