From owner-freebsd-hackers Wed Feb 22 09:52:51 1995 Return-Path: hackers-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.9/8.6.6) id JAA23828 for hackers-outgoing; Wed, 22 Feb 1995 09:52:51 -0800 Received: from cs.weber.edu (cs.weber.edu [137.190.16.16]) by freefall.cdrom.com (8.6.9/8.6.6) with SMTP id JAA23822 for ; Wed, 22 Feb 1995 09:52:48 -0800 Received: by cs.weber.edu (4.1/SMI-4.1.1) id AA24312; Wed, 22 Feb 95 10:46:04 MST From: terry@cs.weber.edu (Terry Lambert) Message-Id: <9502221746.AA24312@cs.weber.edu> Subject: Re: FYI.. To: ugen@netvision.net.il (Ugen J.S.Antsilevich) Date: Wed, 22 Feb 95 10:46:03 MST Cc: hackers@FreeBSD.org, jkh@violet.berkeley.edu In-Reply-To: from "Ugen J.S.Antsilevich" at Feb 22, 95 12:58:44 pm X-Mailer: ELM [version 2.4dev PL52] Sender: hackers-owner@FreeBSD.org Precedence: bulk > Hmm..again completely sick NetBSD "release".. > Tell me, is there *ANY* such release of NetBSD which supports > without any exclusion or stuff at least 70% of hardware for some > architecture? That funny OS just comes out anytime with messages like the > one you sent: SUPPORT FOR > And then: except [ethernet|discs|video(no X,never!!!)|serial|processors|etc..] > Actually i think this is even too small to be called "FYI"..:) > *IMHO* By this argument most "releases" on the net are really Beta distributions anyway. I wouldn't necessarily disagree with this from a commercial viewpoint, but this list is not a list for a commercial product. I'm actually very interested in this port, since a 160 MHz Alpha motherboard is available from DEC Direct for $1170 retail (you should never pay retail!), about the same price as a good P90 motherboard. Admittedly, the $1170 board is PCI based, and so won't work without some effort given the current state of the port. But the port has other value in providing information on 64 bit architectural changes needed in both NetBSD and FreeBSD, as well as providing some cleaned up tools ported to a 64 bit aware kernel. Finally, the comments on ecoff files and getting the symbols makes it clear (to me anyway) that utilities that depend on things like symbols instead of things like procfs are dated and should die a horrible death; the user space utilities other than the developement tools should not be dependent on the object file format used. They can, however, depend on published cross-platform interfaces like procfs. This implies certain file system constructs are mandatory, ie: the mounting of procfs can not be left up to the user, it must be implied by the kernel. I was also interested (but not pleased) by the way they solved the inter-OS file system mounting problem. That the problem exists points out the inadequacies of not isolating the disk subsystem dependencies from the device abstraction, and from not making the per system physical layout of logical partitioning information a seperately interposable layer. The need to move removable media from one machine to another (such as Syquest and magneto-optical cartridges) clearly argues for the ability to interpose the layering on a per device rather than on a per architecture basis. All in all, I think the announcement (which the NetBSD people did *not* make to the FreeBSD lists -- it was echoed by Jordan as an item of general interest) did a lot to make me think about problems for which I did not already have concrete soloutions. Terry Lambert terry@cs.weber.edu --- Any opinions in this posting are my own and not those of my present or previous employers.