Patrick Elliott wrote:
> Well, maybe true. We also might have ended up with a mil-spec system 
> that is stable, required to operate in a wide range of conditions, and 
> clearly documented interfaces.

We might have, but I doubt it.  That would be over-engineering to a 
tremendous degree.

Where such things are necessary, they're built without too much trouble 
beyond tremendous time and cost.

> What we have is unstable, often doesn't 
> operate even in the conditions it supposedly was designed for, and even 
> the clearly documented interfaces, like modem commands, get screwed up 
> by nonstandard interfaces/drivers/control mechanisms, which where 
> designed to work only with specific OSes.

This is pretty much true of every piece of long-lived general-purpose 
software. It's impossible to foresee every eventuality. If your OS 
doesn't support removable media or streaming 3D holograms, you're going 
to have unstable programs that don't work right when you remove the 
media out from under them or stream 3D holograms into their input files.

> In other words, we started with 100 species of machines, none of them 
> alike, and ended up with something that... has almost as many hacks, 
> bugs, design short cuts, and stupid compromises as human DNA (and not 
> because those things "worked" better than the other paths tried). 

But we have that mainly because of age and backward compatibility. It's 
not like "autoconf" is a gem either, and for exactly the same reason.

> If the 
> military built aircraft the way Microsoft, to a large degree, and 
> others, to different degrees, pushed us to produce computers,
> we would be losing billion dollar airplanes once a week, 

They wouldn't cost a billion dollars. They'd cost $1000, and they'd run 
pretty well until you put a trailer hitch on them and tried to tow your 
RV behind. :-)

> Its a logical fallacy to presume that Microsoft was "necessary", any 
> more than just about any other absurd thing that led to the modern world 
> "had to" happen to get here.

Of course.  What do you see as a motivating factor for hardware to get 
standardized beyond someone selling a fairly cutting-edge software 
package not tied to a particular piece of hardware?

 > Even if true, the real question is, "Was
> the price paid to do so *worth* it, given other paths that may have led 
> to the same thing?"

Except they didn't. The world had the chance for 10 or 15 years, and 
there were just as many competing incompatible brands when PC-DOS came 
out as there were ten years earlier.

> PCs imho where inevitable. Had Microsoft not shown up, someone else 
> would have. Eventually, some standard would have appeared. Likely, given 
> the wide us of Unix, it would have been a *nix standard.

I disagree. UNIX puts too many requirements on the hardware for it to 
work at the time.  What we *did* get that was vaguely standard was CP/M, 
which is what PC-DOS was based on.

And, OK, which UNIX? Why do you think it would have been a standard, and 
which one, and would it really have led to there being fewer flavors of 
UNIX to program against? Even now, there's a dozen or more flavors of 
UNIX in current use. We already *have* a "unix standard". You *still* 
need autoconf, and it's broad enough Microsoft managed to implement it 
in Windows (for some meaning of the word "implement").

Blame BSD for that, perhaps?

-- 
   Darren New / San Diego, CA, USA (PST)
     "That's pretty. Where's that?"
          "It's the Age of Channelwood."
     "We should go there on vacation some time."

Post a reply to this message