> On Thu, Dec 06, 2001 at 04:13:51PM +0000, Timm Murray wrote:
> > > Arrrgh. This is silly.
> > > 
> > > The cycle is 
> > > 
> > > Design
> > > Implement
> > > Test
> > > 
> > > Rinse and repeat.
> > 
> > No, that's what they teach you in pea-brained CS classes.
> >
> 
> Well, I never had a CS class.
> 
> But you tell me how logical it is to say
> 
> "I can't state the hypothesis
> until I design and run the experiment"

CS is not a physical science, it is more closely related to mathematics.  
Typically, 
mathematics people might be aiming for a certain goal (through a proof), but 
they may 
spend pages and pages on dead ends before coming upon the right solution.  Then 
somebody cleans it up and writes a paper containing only the useful parts of 
all that 
work.

> 
> Coding is a mechanical process. It *is* the translation of
> a design into a specific implementation.
> 
> The only question is at what level the iteration happens, and
> whether or not you wish to collaborate with others in the process.
> 
> If you are working by yourself, the iteration can be very fine grained,
> with the code being the design documentation, and the design evolving
> with every line of code you write. 

Actually, I don't write code like that, either.  Normally, I write out (usually 
on dead trees) what I want to do, perhaps some notes on the API, and how to fit 
stuff together.  The chicken scratch I scrall in the margins can hardly be 
considered a 
specification, but it isn't "desgin evolving with every line of code" either.

In fact, the Freenet developers as a whole have long since completed something 
similar 
to the above.  Between 0.3 and 0.4, there was a big peer review process that 
went over 
node announcement and ARKs (others?, and those were well-documented over the 
mailing list.  
The specific implementation of those ideas don't have a general spec, they're 
just a 
bunch of code sitting in a CVS tree.

_______________________________________________
Devl mailing list
Devl at freenetproject.org
http://lists.freenetproject.org/mailman/listinfo/devl

Reply via email to