C++ programming (was Newbie Alert: Beginning parallel program ming with Scyld)

Josip Loncaric josip at icase.edu
Tue Oct 22 09:38:39 PDT 2002


On Fri, 2002-10-18 at 22:07, Karen Shaeffer wrote:
> 
> The point is that with every layer of abstraction that you don't have
> control of, your overall system risk goes up. This is the other side of the
> equation. And, as has often been discovered after the fact, fixing system
> level failures that are sitting on top of layer after layer of abstractions
> the development team doesn't control, is often times impossible.

I'd like to second that.  Quick testing of concepts in an advanced
interpreter like Matlab is great; but once one has figured out how the
problem can be solved efficiently, it is time to write carefully tuned
production code.  The pressure to do so is particularly intense when the
problem is computationally difficult, but even simple applications
should be performance optimized if they are used often.  Layering
abstractions upon abstractions for their own sake kills projects.

My own preference is for lean and fast applications.  While writing an
entire browser or e-mail client in interpreted Java may save
programmer's time, it wastes my time.  Today, you virtually need a GHz
machine to read your e-mail, whereas a decade ago machines a hundred
times slower could do this in a snappy manner.  Multiply the time spent
waiting for a fancy e-mail client to re-scan your Inbox by millions of
users and by hundreds of times per day this is done, and you've got a
real time waster.

Sincerely,
Josip

P.S.  C++ is just fine in my book, as long as one avoids getting too
fancy and keeps performance under control.  BTW, those caveats apply to
any language: one can write horrible code even in assembly.

P.P.S.  Sure, a 10 GHz machine will be snappy while reading my mail. 
However, by the time 10 GHz machines are out, mail client software will
get even slower.  Extra features are nice, but I'm almost considering
going back to pine or elm...

-- 
Dr. Josip Loncaric, Research Fellow               mailto:josip at icase.edu
ICASE, Mail Stop 132C           PGP key at http://www.icase.edu./~josip/
NASA Langley Research Center             mailto:j.loncaric at larc.nasa.gov
Hampton, VA 23681-2199, USA    Tel. +1 757 864-2192  Fax +1 757 864-6134




More information about the Beowulf mailing list