From: Steve Lesh (lesh@smoke.brl.mil) Subject: Open response from Bill Jolitz Newsgroups: comp.unix.bsd View: (This is the only article in this thread) | Original Format Date: 1992-05-17 04:42:52 PST Received: from vgr.brl.mil by SMOKE.BRL.MIL id aa05456; 16 May 92 15:35 EDT Received: from cardio.ucsf.EDU by VGR.BRL.MIL id aa06890; 16 May 92 15:19 EDT Received: by cardio.ucsf.EDU (5.61/GSC4.21) id AA06242; Sat, 16 May 92 12:19:18 -0700 Date: Sat, 16 May 92 12:19:18 -0700 From: William Jolitz Message-Id: <9205161919.AA06242@cardio.ucsf.EDU> To: lesh@BRL.MIL, william@ucbvax.berkeley.edu Subject: Re: Some help for the struggling masses Cc: david@ureview.com > Bill and Lynne, > For those of us still struggling with hardware configuration prob- > lems in our attempts to get 386BSD up and running a few non-technical but > perhaps useful ideas: > 1) publish the list of registered configurations which DO work to help > those of us with problems isolate system components which may be causing > trouble; Excellent idea. That was why we included the REGISTRATION form and survey in the first place, and put it into all volumes of the release. For those who registered, we have that information, and it has allowed us to isolate the pattern of problems present. We have found two major items caused most of the troubles: 1) Release confusion and corruption (e.g. using the pre-release version that got around the world before the real release got out, missing documentation, including deleted read.me (no installation procedures), registration (no pointer to contacting us or reporting problems), and manifests files (to cross-check file corruption) and partial distributions (no binary, incomplete source which is impossible to get working); and 2) pervasive compatibility problems, which were expected. We are waiting on approval to release through a more controlled forum to comprehensively deal with item 1, and we are spending long, lonely, and bleary nights/days proving that all the compatibility problems we can find with every recorded problem are fixed with 0.1, currently stuck in release engineering (regression testing). Unfortunately, this is hard work, since many fixes for one configuration may break another working one. Our goal is to make 0.1 a hardy and enhanced version of 0.0. > 2) establish and maintain a list of system components which DON'T > work together with the manifestations of their failure; Yup. Anything that relys on the BIOS (like 8-bit MFM controllers) fits this catagory well. The so-called "" or waiting bug is entirely due to a race condition on controller/drive diagnostics, and is not easily characterizable, because it depended on processor speed, controller type, and drive type/mfr. It was easier to fix than to describe (wait for drive ready after reset, then reprime interrupt enable). I mention this by way of saying that some kinds of bugs are better cured with software than mail. With 0.1, we hope to include a mechanism for distribution of approved patches and additions, alongside unoffical and unapproved patches and additions at distribution sites. We have not done this before, because of difficulties in having a clear and reliable communications channel to work with. Lacking that, we did not wish to add to the confusion present, especially when only erratic portions of our communications might make it back to the 386BSD user. Communications alone has been the biggest weakness. > 3) publish suggestions for least-cost and/or elegant solutions for > component incompatibilities; We'd love to do this, but it's hard to at this stage because it seems to annoy some of the manufacturers when we do so. That's another reason for working from the registration forms -- it's unbiased and authoritative from someone who has the equipment. My suggestion is to work with cooperative groups, be they users or manufacturers or dealers, who want to assist in seeing 386BSD become more solid and real. With the distribution single floppy system, a user can qualify a system, and if it does not work, a dealer can contact the manufacturer, and the manufacturer can contact me directly and get the problem resolved. Until we get 386BSD out of its infantcy, I'm going to field all manner of troubles with it myself from those who can characterize the problem and offer a way to make it accessible to me. Every problem so presented and communicated to me has been fixed and the fix incorporated into 386BSD. It is always difficult to get programmers to behave in a cooperative manner, especially since most talented programmers are individualists. However, with 100,000+ copies in circulation, people seem to be developing a new initiative (despite the inherent inertia). I think 0.1's stability and compatibility will provide them with a very strong motivation to be interested in assisting 386BSD users. > 4) publish a list of console displays which don't necessarily in- > dictate "end of the line" but signify some configuration or installation > procedural problem which may be within the means of the installer > to correct. You mean by this to have a problem isolation guide ("If you see this, try that ") ? Sounds great, I'll see if we can publish such a thing, after 0.1 hits the road, hopefully very soon now. > Finally, (an obviously technical idea beyond my abilities and perhaps > not feasible) would it be possible to write a diagnostic which examined the > hardware configuration of a potential 386BSD user and identified system > components which may be causing trouble? I've been getting some programs from *very good* people to fault isolate problems, and hope to integrate them in future releases as a qualification procedure. > Organized and directed by somebody who knows what they are doing, > this may be an area in which those of us with little or no experience writing > or maintaining UNIX operating system software can make a contribution (and > in the process get an education). Precisely. Be patient, that is exactly the goal. I encourage others to particapate with 386BSD, write about it, and, especially, teach how to use it and make it useful. > Thanks for your efforts in bringing 386BSD into being. I will try > to come up with some cash in the near future to do my little bit in keeping > your energies focused where they can do the greatest good for the greatest > numbers. Contributions are very appreciated and essential to seeing 386BSD reach its goal. If even as little as just 1% of all 386BSD users were to contribute, the project would accelerate tremendously. 386BSD survives entirely on the good will of it's contributors, sponsors, users and enthusiasts. It exists solely to reflect that good will back into this community, to again build even more good will as a result. It lives or dies by this. 386BSD was founded on the notion that people would appreciate being trusted with the ability to participate in achieving the goal of a modern, state-of- the-art, functional, and widely-understood operating system. Trust, honest work and "good will" insure the success of 386BSD. It's that simple. Some might find this an odd, insignificant, naive, or simplistic an objective. Perhaps these values are in short supply these days in our cynical world. But we must not allow ourselves the indulgence of cynicism and bitterness. They sap our strength and keep us from even trying to achieving our grander goals. Don't let this happen. Keep sight of the goal -- it's within reach. Bill. PS. Please post this for me on the network, it's an open response to many who have asked the same questions, and might like an answer. ***NOT PART OF JOLITZ'S REPLY*** Sorry about the delay. This is the first time I've posted and I wanted to wait until I had the time to at least try to do it right. Steven Lesh