Serious investigations into UNIX and Windows
Kevin Lyons
kevin_lyons at ofdeng.com
Wed Oct 27 13:08:51 PDT 2004
Charles Oppermann wrote:
> Yep, he said that, but there wasn't anything to support the statement, other
> than his opinion.
>
Well, what about what his co-engineer has said...
"NT has never been fully refined and there are times when we have had
shutdowns that resulted from NT.”
> However, there are crumbs of evidence to suggest that the fault was in the
> third-party software; mentions of data entry, database software, etc.
Crumbs is about all it is. You are making the point yourself but from a
different angle. a) the official description of the event does not make
sense. b) what does make sense and what the engineers on the project
have said is that NT is the source of the problem and also that the
official description of the events in this case are inaccurate.
> Regardless, that's not central to this discussion. My point is, when an
> application stupidly performs a divide by zero operation, Windows and UNIX
> both handle the exception and terminate the process. I don't think either
> platform helps the application recover from what is a programming mistake.
Yes that is how kernels handle div/0. Therefore, the official
explanation of what happened does not make sense. You are saying the
third party app crashed because it was shit. The engineers on the
project say it was an NT problem. Furthermore DiGiorgio says tongue in
check that it would seem that the divide by zero protection your
calculator has was seemingly not provided on this ships computers.
Obviously it was, therefore he is saying the official explanation is false.
If it was just a case of terminating the application, it seems unlikely
that they would have been down for two hours. 5min would seem more
likely. Wouldn't you also expect some kind of watchdog process. Their
solution is to "retrain the program administrators" to override bad data
fields. Huh? Makes no sense. If it was divide by zero, then do a damn
check and move on. Why leave that in the hands of humans. Again
doesn't make sense,
>
> Disclaimer: I worked on Windows NT during 10 years at Microsoft as a
> developer and manager. In fact, today, I'm wearing my "NT GUI Dev Team"
> shirt. Currently I work on a BSD-based storage networking product for a
> small company. I really like BSD, but having examined both the NT and BSD
> kernels, do not feel that either deserves the reputation they have.
>
> Thanks for the fun exchange!
>
No offense, but I wondered from where the evangelical defense of NT was
coming. I agree, BSD Unix is not perfect, but I haven't found the
perfect OS yet, nor any that come close. And I have tried them!!! boy
have I tried them!!! still healing the scars.
More information about the freebsd-chat
mailing list