I don't mean to top-post -- this is a preface: I hear the slow,
agonising creak of my arguments being demolished, one by one, by a
technician .. and yet .. and yet .. I am too much a romantic to give up
altogether (it's nothing I'm proud of). Nevertheless,
Joel said:
True, although my opinion is that for complicated test instruments such as
analyzers & scopes, how well the software works has much more to do with the
programmers than with the underlying OS.
That's difficult to argue against.
That is generally indicative of bad programmers, not bad OS design. Yes,
Windows (and every other commercial OS out there) has bugs (tons of 'em), but
for every PC that crashes probably only 1 in 1000 if not 10000 does so due to
an OS bug rather than an application or 3rd-party driver bug.
And that's also quite true. NT is rather a good design at the core, and
most crashes do indeed come from drivers.
There's still a load of horrible design in the things that go around
it: for instance, can there be any reasonable explanation or excuse for
the way USB driver installation is handled in Windows XP? And however
beautiful the heart, it's wrapped in a pig's skin. It feels to me like
it was made by people with no taste and no concern.
If you keep your instrument off the network -- or within a well-firewalled
intranet -- realistically the security patches are highly optional. ANY
instrument -- regardless of OS -- that's sitting directly on the Internet
needs to be kept up-to-date security patch-wise.
This is really only an issue because older instruments didn't *have* TCP/IP
connectivity...
You are completely correct. I cower corrected.
In my experience, QNX, Linux, Windows, and VxWorks all take roughly comparable
(same order of magnitude) times to boot, although I've seen people pay a lot
more attention to getting QNX and VxWorks to boot *quickly*. Of all 4, Linux
is probably the slowest, since it doesn't have the fancy background service
that Windows does that specifically goes around trying to impove your boot
times (but still has just as many services to load).
Once again, a well-made point.
My life-meter is getting dangerously low ..
Umm... have you hung out on any of the groups where people discuss trying to
clone old instruments' EEPROMs to try to get a dead one working? Or how to
read some proprietary disk imagine format from a 20-year-old instrument? I'd
much rather take my chances with a common PC platform -- even if the
manufacturer has added proprietary disk cases or something -- than with true
fully-custom platforms.
Ouch! Ouch! It hurts!
I was really thinking here of the "golden days", which I admittedly
don't remember myself, when (so I have heard) scope repair was a
routine task in a lab.
But those were fully analog scopes, and at any rate it's rarely wise to
talk about how wonderful the old days were, because mostly, they
weren't.
That's naive... I know workmen by their results; I'm not too concerned with
what tools they use. I've seen plenty of guys with far better tools produce
inferior work to someone with low-end tools just because the later was more
experienced, concerned about quality, whatever.
Now, on this one, I do not bend. There's wisdom in that saying; I
didn't invent it. And it remains that if a craftsman uses cheap, flimsy
tools, he will usually not be as good as one who uses good tools,
because a good craftsman will tend to choose good tools -- not
necessarily fancy and expensive ones, but always the *right* ones.
I agree that results must be the measure of everything (well, to a
point). But sometimes I must judge a worker before I've seen results,
and if he is unconcerned about his tools, he is either third-rate or
(much rarer) a craftsman of unusual skill.
There's a good market for that sort of instrument, and they are quite useful,
but there's always going to be demand for an instrument with its own box, its
own custom controls (knobs and buttons), etc.
I didn't mean to say there wasn't -- I'm a tremendous fan of knobs and
buttons (and flashing lights), and the more the better.
You have done plenty of complaining
Well, to be honest, I was in a bad mood, and probably shouldn't have
been posting. I'm better today.
and made some good points, but let me
query you -- what OS would you *prefer* manufacturers of test equipment to
use?
This is where my inexcusable romanticism and personal prejudices come
to the fore.
I would prefer that they use any OS at all, even Windows, as long as
the instrument itself has a solid feel -- as long as it feels like an
instrument. I want it to be something that isn't cheap -- something
made by people who think, and who care.
But Windows is a special problem, because I regard it as the triumph of
the cheap and second-rate. I don't want my bench infected by its
shoddy, amateurish presence (my own presence fills that quota and then
some). Perhaps I should relax and be more objective -- but it's hard
for an inveterate perfectionist.
Well, sir, if you didn't leave me in a heap on the floor, I at least
require a trip to First Aid. Good work, and good points all.
cheers --mpa
