jon@runxtsa.runx.oz.au (Jonathon Seymour) (04/10/90)
In article <28655@cup.portal.com>, Will E Estes writes: > What this comes down to is a high-risk gamble by Microsoft. On the upside, > Microsoft hopes to make OS/2 look so good that people will simply abandon > DOS and make a radical shift to OS/2. For some of us this will work. However, > by and large, I think this strategy misunderstands human nature. People do > not like revolutions; they like evolution. Had OS/2 given people a way to > run *all* of their DOS applications and third-party hardware under OS/2 > it would have had a good chance of succeeding. > > ... > > It's really a shame because I happen to think that OS/2 is a *hot* operating > system, and I would love to see it on everyone's desktop. It would certainly > solve a lot of constant hassles in a lot of DOS users' lives. But Microsoft > is certainly not making it easy for us to embrace OS/2 with open arms. > > Will (sun!portal!cup.portal.com!Will) > It would have been a misunderstanding of human nature to drop DOS support altogether. But Microsoft did not do that. If OS/2 represents revolution, rather than evolution, then long live the revolution! After a point operating systems stop evolving and then it becomes a mindless struggle to keep them useable (eg. MVS). With the advent of multi-tasking interfaces, networks and more sophisticated CPU's, MS-DOS was fast approaching that point. Sure, these things can and have been done with MS-DOS, but in each case they have been achieved with minimal architectural support. Lest we think that architectural support is a trifling issue, consider how I/O is handled by UNIX and MVS. In UNIX virtually all I/O whether to file, to terminal or between processes is done with the same mechanism - the file descriptor. In MVS there are not just different mechanisms for each type of device (terminal or disk), but there are also different mechanisms for data which resides on the same type of device (sequential access methods vs. VSAM, for example). All because MVS does not provide architectural support for the concept of a file (as it is understood in its UNIX context). Sure, the functionality is there, but at what cost! Microsoft's strategy of providing a Family API for all new applications seems to be a sensible one - extending DOS to be compatible with OS/2. It is certainly more sensible, in the long term, than doing the reverse. Instead of taunting Microsoft we should be giving them a gentle pat on the back for taking some relatively courageous and innovative decisions in the design of OS/2. As people with an interest in computing technology it is surely our duty to leave the JCL's of today dead and buried. jon.