sater@cs.vu.nl (Hans van Staveren) (11/22/90)
Recent developments with a workstation supplier that shall, for the moment, remain unidentified, have made us think about the following: Is it reasonable to assume that if you buy computer hardware from a supplier that the information needed to program that hardware should be available to you? Information like the address of the Ethernet chip on the board, details about memory management and the like, in short everything you need to know to write your own operating system. What is the position of the current major vendors? If they all make the same fuss about it as the one we are dealing with now the research into operating systems can only test real systems on strange hardware. That can hardly benefit the progress of operating system design. Hans van Staveren Vrije Universiteit Amsterdam, Holland The Amoeba Group
henry@zoo.toronto.edu (Henry Spencer) (11/24/90)
In article <8319@star.cs.vu.nl> sater@cs.vu.nl (Hans van Staveren) writes: >Is it reasonable to assume that if you buy computer hardware from a >supplier that the information needed to program that hardware should be >available to you? Is it reasonable? Yes. Is it realistic? Not any more. Buy from MIPS. They still seem to ship complete documentation. Avoid Sun in particular; they never have. -- "I'm not sure it's possible | Henry Spencer at U of Toronto Zoology to explain how X works." | henry@zoo.toronto.edu utzoo!henry
john@newave.UUCP (John A. Weeks III) (12/02/90)
In article <8319@star.cs.vu.nl> sater@cs.vu.nl (Hans van Staveren) writes: > Is it reasonable to assume that if you buy computer hardware from a > supplier that the information needed to program that hardware should be > available to you? NO, this is not a reasonable assumption. Many systems are sold to perform a specific task and the details of the implementation might be a strictly held secret. Another reason for keeping the implementation details secret is to allow them to make changes or upgrades to the system without fear of breaking existing applications. This is one of those cases where common sense != business sense. I recently worked on a contract for a computer systems organization that supplied only the K&R C book for documentation along with a short note about how most of the standard library was not implemented because the O/S did not support file type I/O. By hacking and poking around through the machine, I did manage to discover a lot of neat things...but in the end, all I really accomplished was to write a bunch of code that will never work past the next operating system upgrade. Needless to say, they kept most of their secrets, but it was not a very productive way of doing things. > Information like the address of the Ethernet chip on the board, details > about memory management and the like, in short everything you need to > know to write your own operating system. I once proposed building a very specialized machine that needed quite a bit of CPU horsepower. To save on hardware development costs, I proposed using the main CPU board from a SUN 3/50 (this was several years ago). All we would have to do would be to add a specialized I/O circuit. Upon approaching Sun with our ideas, we were told by Sun that the internals of their machines were off-limits to developers and they would not provide (sell) any info other than what was documented in their UNIX manuals. So I used a 386 clone motherboard when they became available. -john- -- =============================================================================== John A. Weeks III (612) 942-6969 john@newave.mn.org NeWave Communications ...uunet!rosevax!tcnet!wd0gol!newave!john ===============================================================================
hrubin@pop.stat.purdue.edu (Herman Rubin) (12/02/90)
In article <531@newave.UUCP>, john@newave.UUCP (John A. Weeks III) writes: > In article <8319@star.cs.vu.nl> sater@cs.vu.nl (Hans van Staveren) writes: > > Is it reasonable to assume that if you buy computer hardware from a > > supplier that the information needed to program that hardware should be > > available to you? > > NO, this is not a reasonable assumption. Many systems are sold to perform > a specific task and the details of the implementation might be a strictly > held secret. Another reason for keeping the implementation details secret > is to allow them to make changes or upgrades to the system without fear of > breaking existing applications. > > This is one of those cases where common sense != business sense. It is possible that if some hardware is being used only for a communications channel, or display purposes, or to run a rigid office system, or a specific type of automation, this may be the case. POSSIBLY this may be the case for a specialized military computer, or in a situation where a programmer with brains is not going to be available. But if one even wants to do text-processing with a sufficiently unusual character set, or even slightly unusual numerical procedures, this is not the case. I can see situations where access to the operating system can lead to a breach of security, but this is not the same as not allowing people to make intelligent use of the machine. -- Herman Rubin, Dept. of Statistics, Purdue Univ., West Lafayette IN47907-1399 Phone: (317)494-6054 hrubin@l.cc.purdue.edu (Internet, bitnet) {purdue,pur-ee}!l.cc!hrubin(UUCP)
ge@wn3.sci.kun.nl (Ge' Weijers) (12/03/90)
.In article <531@newave.UUCP>, john@newave.UUCP (John A. Weeks III) writes: .> In article <8319@star.cs.vu.nl> sater@cs.vu.nl (Hans van Staveren) writes: .> > Is it reasonable to assume that if you buy computer hardware from a .> > supplier that the information needed to program that hardware should be .> > available to you? .> .> NO, this is not a reasonable assumption. Many systems are sold to perform .> a specific task and the details of the implementation might be a strictly .> held secret. Another reason for keeping the implementation details secret .> is to allow them to make changes or upgrades to the system without fear of .> breaking existing applications. There is only one way to solve this problem: vote with your budgets. No specs, no sale. A machine without specs is not very useful for research, so universities and research centers should buy elsewhere and of course also explain to students why they buy elsewhere, so that those can later avoid purchasing from companies with too many useless trade secrets. In casu SUN: I wonder whether they will survive the loss of the research/university market. Ge' Weijers -- Ge' Weijers Internet/UUCP: ge@cs.kun.nl Faculty of Mathematics and Computer Science, (uunet.uu.net!cs.kun.nl!ge) University of Nijmegen, Toernooiveld 1 6525 ED Nijmegen, the Netherlands tel. +3180652483 (UTC-2)
jack@cwi.nl (Jack Jansen) (12/03/90)
ge@wn3.sci.kun.nl (Ge' Weijers) writes: >.> In article <8319@star.cs.vu.nl> sater@cs.vu.nl (Hans van Staveren) writes: >.> > Is it reasonable to assume that if you buy computer hardware from a >.> > supplier that the information needed to program that hardware should be >.> > available to you? >There is only one way to solve this problem: vote with your budgets. >No specs, no sale. And then wait for the hassle. The only way to be sure that you get sources or specs is to have it in the sales contract *in writing*. Salesman waving hands and telling you everything will be fine are not to be trusted. Even then, I know of various vendors that will try their best to get out of the obligation, and do everything in their power to postpone the delivery of sources or specs [no names, just in case our experiences aren't representative]. Credit where credit is due: Digital is the one exception that I know of. They do take a looooooong time before delivering sources (mainly due to zillions of burocrats in between our contact here in Holland and the people who actually have to sign things and send tapes in the US, I've been told), but when they do you get the latest stuff, even though that might be newer than what you paid for. Moreover, what you get is as complete as can be: standalone stuff etc all included. (with the exception of things that they can't distribute because they aren't theirs, like the mips compilers, etc) Also, note that many vendors aren't interested in this sort of customers anymore: the research market isn't that big, and most of that market will be satisfied as soon as they can write a scsi or vme controller to hook up their experiment. The days that many people were building custom hardware to plug into their pdp-11 are gone.... -- -- Een volk dat voor tirannen zwicht | Oral: Jack Jansen zal meer dan lijf en goed verliezen | Internet: jack@cwi.nl dan dooft het licht | Uucp: hp4nl!cwi.nl!jack -- -- Een volk dat voor tirannen zwicht | Oral: Jack Jansen zal meer dan lijf en goed verliezen | Internet: jack@cwi.nl dan dooft het licht | Uucp: hp4nl!cwi.nl!jack