peter@thirdi.UUCP (Peter Rowell) (05/25/88)
Speaking as a software developer who wants his code to run on as many machines as possible (or to port easily to many machines), the whole thing with OSF just doesn't matter. You *know* that they will be POSIX conforming or else they won't be able to bid on those nice, fat government contracts. If you write code that uses a local feature put in by a manufacturer, *you* are the one guilty of non-portable code. This goes just as much for a "special" feature in BSD or System V as it does for OSFix (or whatever they're going to call it). You might think of these "features" as a type of drug: they have attractive aspects (I guess) and they seem to be cheap (at first), but you *know* better than to use them becaue Mom said "Beware of proprietary features, because they'll rot your brain and make your code pure hell to port!" Mom knew what she was talking about! ---------------------------------------------------------------------- Peter Rowell "Yes, Mother....." Third Eye Software, Inc. (415) 321-0967 Menlo Park, CA 94205 ...!pyramid!thirdi!peter
bak@csd-v.UUCP (Bruce) (05/27/88)
In article <415@thirdi.UUCP> peter@thirdi.UUCP (Peter Rowell) writes: > >Speaking as a software developer who wants his code to run on as many >machines as possible (or to port easily to many machines), the whole > >You *know* that they will be POSIX conforming or else they won't be >able to bid on those nice, fat government contracts. If you write code >that uses a local feature put in by a manufacturer, *you* are the one >guilty of non-portable code. This goes just as much for a "special" Yes, this is an excellent point. When I first started playing around with my AT in 1984, the os that I got with it was PC DOS 3.1. So I naively wrote some code which used the DOS calls promiscuously and when I took a diskette to work to show some friends what I had wrought I quickly discovered why the os standard was *NOT* simply DOS, but rather DOS 2.0, the lowest common subset which supported tree structured directories. I thereafter never used any system services which were not in "common" DOS (2.0). A point I'm sure that is already well known by developers of all *IX based software. -- Bruce Kern | uunet!swlabs!csd-v!bak Computer Systems Design | 1-203-270-0399 29 High Rock Rd., Sandy Hook, Ct. 06482 | This space for rent.
bak@csd-v.UUCP (Bruce) (05/28/88)
>In article <415@thirdi.UUCP> peter@thirdi.UUCP (Peter Rowell) writes: > >Speaking as a software developer who wants his code to run on as many >machines as possible (or to port easily to many machines), the whole > >You *know* that they will be POSIX conforming or else they won't be >able to bid on those nice, fat government contracts. If you write code >that uses a local feature put in by a manufacturer, *you* are the one >guilty of non-portable code. This goes just as much for a "special" Yes, this is an excellent point. When I first started playing around with my AT in 1984, the os that I got with it was PC DOS 3.1. So I naively wrote some code which used the DOS calls promiscuously and when I took a diskette to work to show some friends what I had wrought I quickly discovered why the os standard was *NOT* simply DOS, but rather DOS 2.0, the lowest common subset which supported tree structured directories. I thereafter never used any system services which were not in "common" DOS (2.0). A point I'm sure that is already well known by developers of all *IX based software. -- Bruce Kern | uunet!swlabs!csd-v!bak Computer Systems Design | 1-203-270-0399 29 High Rock Rd., Sandy Hook, Ct. 06482 | This space for rent.