ftp.nice.ch/peanuts/GeneralData/Usenet/news/1989/CSN-89.tar.gz#/comp-sys-next/1989/Jan-Apr/Development

This is Development in view mode; [Up]


Date: Sun 09-Feb-1989 14:43:25 From: Unknown Subject: Development Dear NeXT, Could someone at NeXT please send me the "Developers info package" Thank you. And concerning source code availability. I have been working with computers for a few years. My first experience was as an operator of an IBM S/34. I was then taught to "program" the thing, which involved writing code in RPG 2, using the command language and some pretty powerful utilities e.g. for file sorting and data entry. It was a small microcosmos for manipulating files of records of fields. We had good documentation and all told it was not bad for what we were doing. Occasional bugs had usually been encountered by someone else before and the next release of the software had it fixed. I would like to focus on two relevant points. 1. We had no source code access. The documentation was very good and we weren't really doing anything "new", just shuffling records. So we did not need source code. 2. The people at IBM were constantly improving the system, adding new features and fixing bugs. There is NEVER a final version of system software. My next project was absolutely different. I adapted the Acorn/BBC computer for use with the national languages of Iceland, Denmark, Norway and Sweden. This was a totally different story. Here I had to dig into the operating system and (with help) managed to change the keyboard layout, screen font and printer driver. Due to lack of documentation for the parts of the OS that I was working on, I actually had to go to Acorn in Cambridge, England. The Acorn OS for the BBC computer is very brilliant. There are hooks in the OS to intercept almost anything and you can direct it to access extra facilities in auxiliary ROMs. I really did not have to do to much work. I had to: Get access to the documentation (what can I do and how should I do it), Get relevant source code (rewrite the keyboard driver), Get help from someone who knew 6502 assembly language (I got a lot of help with the coding). So what happens? There is a minor problem in then OS, one that someone not doing an Icelandic keyboard driver will never notice. I look at the source and there is a comment "If we can spare a few bytes in the ROM and some cycles in the interrupt we will do this differently". They never did and over the years it has cost us months of work to write around. To restate the relevant points as in the previous example. 1. At the time Acorn had not released documentation on the parts of the OS that I was working on. I was doing something that had not been done on their machine before. Access to the source code was thus essential to the completion of this project. 2. Operating systems should constantly be improved by adding new features and removing bugs. There will always be developers, who for some obscure reason run into bugs or are doing things that are incompatible with the way your operating system is written. They must be able to modify your system to test the validity of potential fixes. THEY are better off fixing your code than having to write around it. It is very much a question of productivity. If you supply source code, then the users of your system that are developing truly innovative applications will excel. They will be able to build on the advanced technology of your system. They will be able to compete favorably with developers doing similar work on other systems. Some developers are constantly trying to optimize the use of human and computer resources. They are debugging systems at various levels. I would like to say that for these people management problems are just like any other bugs. So they ask you to supply the source code of your system. If you don't, they will have to revaluate their options. You are running a race. If developers can do new things in a more productive way on your system, they will help you win that race. Gudmundur Ragnar Gudmundsson P.O.Box 156 121 Reykjavik Iceland ragnar@kopasker.is >From: wtm@neoucom.UUCP (Bill Mayhew)
Date: Sun 10-Feb-1989 21:08:02 From: Unknown Subject: Re: Development I would prefer source be available if possible. It might be fair to attach a fairly high monetary cost to furnishing source, thus assuring that only companies/people who really, really, honestly need it go to the effort to get source. There are some advantages to keeping source proprietary. The Apple Macintosh systems are good examples. With the Mac. it is often a pain in the neck to code in the manner Apple prescribes. There is a strong temptation to say "if I knew where to pointer to blah-blah were, I could just go in and modify the pointer directly". Unfortuantely, the Mac. ROMs have undergone fairly frequent revisions, and the internal structure of the current 256K ROMs is vastly different from the original 64K devices. Unfortunately, witholding source hasn't stopped industrious coders from reverse- engineering the salient sections of code and hacking in. Even fairly main-line applications, even some from Claris (Apple's software spin-off company), have been guilty of poking around in the O/S and have been burned when new ROMs appeared. The same situation is analogous to different revisions of the Next kernel. One hopes that the cube has enough horsepower that most, if not all, of the time going through the high-level interface to system services will not be overly penalizing, unlike the original Macs. which were pretty s-l-o-w. It wouldn't be too surprising that people writing networking stuff would need to have access to source to really understand the kernel data structrues and software interfaces. I can see both sides of the argument. But.. I believe source should be available at a price, if you really need it.... There are quite a few times when I've been very glad that we have a source license for our bsd relase. --Bill wtm@impulse.UUCP ...!lll-winken!scooter!neoucom!impulse!wtm >From: kean@mist.cs.orst.edu (Kean Stump)

These are the contents of the former NiCE NeXT User Group NeXTSTEP/OpenStep software archive, currently hosted by Marcel Waldvogel and Netfuture.ch.