[rescue] slow java code...an example of why it might be happening

[Jonathan C. Patschke]

On Tue, 11 Feb 2003, Dave McGuire wrote:

> His brain was trained to build an object, then "close the door" on
> that object and completely forget how it works...as long as you
> remember its interface, you can just use it anytime you want.

That sound you just heard was a thousand CS professors' brains
twitching.  You've just described OOP as it's -taught- in college.
"Create it and forget it!  Now you've taught the computer how to do xyz,
so you don't have to remember how."  It's a good idea up to a point, but
the minute you throw any resource management into the picture,
everything starts moving in slow motion.

>    There's a such thing as a programming language whose conceptual level
> is TOO HIGH.  This is a big problem with inexperienced programmers.

Yep.  And more and more of them are being told "instruction set doesn't
matter" by the industry and "unsafe languages are bad" by academia and
"out GC code can manage resources far more efficiently than you'll ever
be able to" by the language vendors.  Is it any wonder that most of the
people who've started writing code in the last 6 years can totally bury
an incredibly fast computer with the OOP equivalent of bogosorting or
busywaiting?

> Most of the horribly slow and bloated C++ code that I've looked through
> suffers from problems very much like the one I described above, and I
> suspect all the commercial farmed-out-to-India bloatware that you guys
> are bitching about is like this as well.

Yes.  This pretty-much sums up HP's experience with outsourced software
development (to India), as related to me by SWMBO's parents, both of
whom worked in HP Labs during the late 90s.

-- 
Jonathan Patschke  *)  "It's not about who's right and who's wrong...
Thorndale, TX      (*   it's about who works for the government and who
                   *)   doesn't."                        --Dave McGuire