RE: Who is a Programmer? (was RE: WDSC and Linux)

> From: Michael Naughton
>
> I guess my question is one of degrees: do I really have to know how to
> write an operating system to be qualified to write an invoice processing
> program?

We've beaten this horse pretty severely, and a lot of my initial rancor was
in response to Aaron's blanket assertion that VS.NET is better than Java
because it's "easier".

But to respond to this particular comment, no you don't NEED to know how to
write an operating system. It is my assertion, however, that the more you
know about how an operating system works the better your chance of debugging
a program, any program.  And the farther your tools take you away from that
level, the farther you are from the "bare metal", the less likely it is that
you will be able to debug something when it breaks.

Now, one real issue is what constitutes "bare metal".  In the olden days, it
was the microprocessor and the machine code.  With OS/400, I believe that at
least 99% of the time I can treat base OS/400 as "bare metal", in that
problems in my program are likely to be from my code, not OS/400.  And it is
this confidence that allows me to be able to debug OS/400 programs at the
HLL level.

With servlet/JSP logic running on the iSeries I have pretty close to that
same level of confidence.  It's not 100%, but Blair Wyman and his colleagues
do one heck of a job keeping us insulated from JVM errors (although I'm
running into one right now, but that's a different topic for a different
day).

Windows operating systems don't work this way.  Errors are often spurious
and inconsistent, and it's much harder to determine why a specific line of
code in a program is causing a certain error.  If I'm running VS.NET, it
means my server is running on a Windows box, and all that wonderful
middleware is coming out of Redmond.  That scares me a little.  No, a lot.

And that's why I don't like to run anything mission critical on Windows.
Simply because I CAN'T debug down to the bare metal.  Back in the bad old
days of Windows 3.1, we could get close, but since the advent of NT and now
W2K and XP, there's really no way for an application programmer to really
know what's going on under the covers.  And if I can't debug it, I sure
don't want my work to depend on it.

Joe