[OT?] Best Practices for Scientific Computing

Eray Ozkural (exa) erayo at cs.bilkent.edu.tr
Tue Aug 14 14:59:59 PDT 2001

Hash: SHA1

On Tuesday 14 August 2001 11:10 pm, Jakob Østergaard wrote:
> Ok, I suppose you don't know about the first worm that took down a
> considerable portion of the Internet - the Morris worm.

:) Yes, I know that story. My point was that today POSIX implementations seem 
to have more robust network software. I think you could make that point, and 
one could also say that it's in the tradition of UNIX and Internet to achieve 
robust systems by following good software practice. (ie open standards/code).

For the specific discussion one might point out better performance and 
reliability which are also important for scientific work.

I cannot emphasize the importance of the tradition of "openness"  enough: in 
research especially you should be able to have a tight grip on the code that 
you are running. Having alternatives is a good thing, too.

I believe I mentioned the advantage coming from the availability of tools on 
UNIX. This is an important point I guess. For software development, UNIX is a 
lot more friendly environment.

On my debian box, I find pretty much everything that I need to write complex 
software. That's a very important point and this means that research division 
will save both time and money.

You can make the point that windows software is generally _not_ geared 
towards high-end research/programming work. And one could say that there is a 
great tradition of research/programming on UNIX side. That is beyond the 
differences in the OS. It's a matter of tradition; one readily uses a lot of 
concrete experience the moment he gets a shell prompt and starts writing a 

Furthermore, the new GUI environments in UNIX are definitely superb. I had 
never thought that a good desktop could improve my performance, however I 
must admit that the new version of KDE is a very solid UI, and it does aid 
the developer in many ways.

When you're on windows, you are almost completely helpless when you want to 
do something slightly complex: you'll have to rewrite the whole world or 
spill money in obscure companies to get the most basic jobs done, you'll 
suffer from the lack of even most basic documentation, etc.

I'm not arguing for the quality of particular codes, but it seems that we all 
benefit from the underlying philosophy: to get things to work well. :)


- -- 
Eray Ozkural (exa) <erayo at cs.bilkent.edu.tr>
Comp. Sci. Dept., Bilkent University, Ankara
www: http://www.cs.bilkent.edu.tr/~erayo
GPG public key fingerprint: 360C 852F 88B0 A745 F31B  EA0F 7C07 AE16 874D 539C

Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org


More information about the Beowulf mailing list