winscape 2276

I'm not necessarily talking about throughput. I am talking about me being in charge of the programming languages I use on my hardware- not the vendor.

But I don't get a chance to fix the compiler when its broken- I have to hack around the vendor's errors- if doing so is even possible.

Oh please. I didn't know this was stand up comedy night.

I don't have a good solution for memory protection in a flat model OS on hardware without memory management resources. There isn't a good solution in that situation- except perhaps for something like the Lisp Machine where you have a very powerful & efficient HLL as the only interface to the hardware. Even then, as flexible & extensible as Lisp is, it still doesn't give you the flexiblity to use whatever compiler you want that meets the hardware & OS ABI.

MS-DOS was crap and remains crap. Yes, it was the only reasonable OS candidate for some time, particularly in the low-end market, but that isn't due to its design quality, but to the lack of effective compebreastion. Xenix and OS-2 were compebreastors in the 386-486 space, but by that point IBM had already given up competing and presumably Microsoft chose to go with DOS because the programmers they had at the time couldn't walk and chew gum at the same time so Xenix wasn't understandable. I remember MS-DOS 4- it was buggy as hell, it didn't perform well & it was unstable & it was insecure.

As far as "goals", its not clear to me DOS had any- other than a vague resemblance to somebody's memory of an ancient Unix. Since its principal function was to act as a program loader for software that took complete control of the machine's hardware, there isn't much of an opportunity for any sort of design goals to apply.

Gregm