Slashdot Mirror
25 Years After DOS - Lessons for Linux?
E IS mC(Square) writes "Microsoft is planning to celebrate 25 years of DOS. An article at ReallyLinux discusses what lessons Linux can learn from the history of DOS. The article begins with 'What can the Linux world learn from Microsoft's past 25 years of unique experiences and domination?', and ends with 'Only question now is not if but when will Linux become the number one OS on earth?'" From the article: "First, we must admit openly once and for all that the 'best solution' is not always the 'most used solution.' There are few who would be foolish enough to argue that back in 1981 PC-DOS was the best solution. There were obviously a number of choices. PC-DOS was the least robust, the most temperamental, and arguably not very compatible with the IBM hardware and BIOS it was sold to work on. Yet, somewhat like the odd but obvious dominance of the VHS over BETA, this simple, cheap OS stole the show."
DEC guy?????
Your refering to Gary Kildall and thus to DR (Digital Research) of CP/M, and GEM fame, not to DEC (Digital Equipment Corporation) of Vax, VMS, and PDP-11 fame.
And yes I'm that old. I do remember when all the above was "hot stuff."
"Oh drat these computers, they're so naughty and so complex, I could pinch them." --Marvin the Martian
Unix has always been used for servers.
Believe it or not, there was a day and time when Unix security was considered a bad joke, and Unix machines were academic or research boxes for the most part. That ended with the entry of Unix powerhouses like Sun, but there was an earlier era. Technically you are right, because in the bare beginning Unix was essentially a time-sharing system with users connected by dumb terminals, and Unix was ONLY a server OS.
It's erroneous to call Linux 'the latest version of Unix.' The BSD OSes are direct decendents (through layers of evolutiona and re-write which excised all the code that 'evil' entities now 'own' and wield like a weapon, of course). Linux is a clone, similar to Coherent, QNX, or OS-9.
Here's SuSE for free:
...
ftp://ftp.suse.com/pub/suse/i386/
You can get the 9.2 Pro install disks, the 9.3 Live CD/DVDs - heck I have the Enterprise Server 9.0 disks here on my desk on CD-R so I must have found them out there somewhere too
With SuSE the cost is for support, not for the actual OS (although they may charge a nominal fee for the retail release in the pretty box with included media.)
And yes, if it weren't for the damn internally used web apps using ActiveX (or IE only features) I would be totally converted over at work.
Glonoinha the MebiByte Slayer
IBM's $39.95 DOS while CP/M was $450 and UCSD p-System was $550. http://pcworld.about.com/magazine/1908p133id52503. htm
http://en.wikipedia.org/wiki/Signature_bloc
First, while IBM had a full licence deal to use Windows 3.1 (a bit remaining from the whole OS2/NT partnership), they made no real effort to make it work well inside their fancy 32bit OS (starting Windows programs resulted in a copy of Windows 3.1 actually being booted up just for that program).
This isn't correct, or rather, it's not an accurate representation of the effort IBM made with Windows for their WinOS2 subsystem.
IBM had access to the Windows source from Microsoft as part of the deal they cut during the breakup. In order to get it to run properly, they made some changes to the WinOS2 subsystem to allow it to run as a DPMI client under their new MVDM (Multiple Virtual DOS Machine) subsystem,they recompiled the code with Watcom's C compiler to improve performance, and they also redesigned the Windows video driver layer to allow a WinOS2 session to poke a hole in OS/2's native PM (Presentation Manager) desktop and display that WinOS2 session alongside the rest of the screen (which was controlled by PM).
The end result was called Seamless Windows, and was both fascinating in its flexibity and disconcerting in its mixing of two window APIs and two sets of Window frames and mouse cursors on the same desktop.
Not only did IBM tweak the video subsystem, but networking, sound, and other elements of the virtualized Windows environment were allowed to use the OS/2 networking, sound, and mouse services, resulting in a hybrid that ran Windows software quite nicely without having to have direct access to any of that hardware (or to use any Windows or DOS drivers).
The WinOS2 subsystem in OS/2 2.0 only supported Windows 3.0 programs (note that Windows 3.1 had been released in APril 1992, roughly the same time that OS/2 2.0 was finally released as a General Availability product), but OS/2 2.1 corrected that in May of 1993, and the so-called emulation of Windows 3.1 was so good between the 2.1 release and the release of Windows 95 that many software vendors saw no real point in supporting OS/2's own native API, and Microsoft chose to respond to this threat by creating over a dozen different "Win32S.dll" additions to the Windows 3.1 API to make Windows a moving target that IBM couldn't possibly keep up with.
The care taken for supporting old DOS programs (which they didn't need Microsoft's help for) was even worse - while Windows 95 needed tweaking options too, OS/2 presented users with a huge checklist that had to have been literally copied straight from the constant names in the C header file (the option names even included the underscore). The options where so badly labeled that even an expert had a hard time figuring out what each option did, let alone what option should be used to get a program to run.
This is total nonsense. The options presented for a VDM were numerous, that is true, but that's simply a reflection of the tremendous amount of flexibility that IBM designed into their MVDM subsystem (a subsystem which has survived almost unchanged though Warp 4 to eComStation today). The options were (and are) clearly labelled, had fairly extensive online help, and were quite clear to anyone familiar with the terminology and options that were present in a copy of actual DOS.
Think of a Windows 3.1 PIF file on steroids.
I'm saying this as a DOS user from 1988 through 1992 who switched to OS/2 2.0 in 1992 from a combination MS-DOS 5.0 and Windows 3.1 environment for the main reason of running multiple virtual DOS machines for using my DOS software collection. I know the OS/2 VDM subsystem inside and out from a user perspective, and it was *trivial* for a knowledgable DOS user to master quickly.
DOS machines under OS/2, by default, used a virtual DOS kernel, not a real DOS kernel. That means they used an interface which looked like the real DOS interrupt interface, but which actually provided a link to OS/2's own system services. Because of this, a DOS program could usually use things like the mouse, soundcard, and networkin
Mainframe/UNIX Bit Twiddler and long time Windows/Linux Hobbyist.
The Theorem Theorem: If If, Then Then.