Squeak-dev Digest, Vol 22, Issue 20

Blake blake at kingdomrpg.com
Sat Oct 16 09:19:04 UTC 2004


On Fri, 15 Oct 2004 13:33:44 -0700, Rick McGeer <rick at mcgeer.com> wrote:

> It was the observation that running a tight cycle of design, code, and  
> test was far more productive that led me to question why, exactly, this  
> was the case.
>
> Which led to the insight about dynamic artifacts, etc.

You know, I'm all for trial-and-error. But there's a distinct rush  
associated with "getting it right the first time." There's also a rush  
associated with turning code into machine language and sending it off with  
a big FINISHED on it.

> A better question is WHY we believe that writing papers (as opposed to  
> code) is design.

The same reason we believe that supply lines and troop movements is war,  
as opposed to just a bunch of guys standing around shooting at each other.

>  A cynic would argue it's because incompetents want designs explained to  
> them in language they can understand.

Wait, if I'm doing the "writing papers" AND doing the coding, am I  
competent when I write the papers and code, but not when I read the  
papers?<s>

> A nicer person would argue that our "software engineering" (hate that  
> phrase) techniques come from the bad old days of time-shared batch, when  
> a tight design-code-test cycle was difficult....

Ha! There wasn't much tighter design-code-test than the bad old days, when  
one of the aforementioned batch sat down with a prorgam listing, an  
obscure error message, a cup of coffee and cigarette to try to figure out  
what had gone wrong. Or maybe with a core dump or a batch of results that  
were slightly off.

It was tight because it was very often one person. He was designer, coder,  
tester--and if you think the "incompetents" today are bad, imagine 30-40  
years ago, when computers were the stuff of sci-fi.

But in some ways, this also made it easier: Your environment was static,  
except for occasionally crashing for 3-4 hours at a stretch. You were  
charged with very specific results, and you were rewarded for getting  
those results correct and (to a lesser extent) without running up the CPU  
time. You did not have to be a graphic artist worried about UI design. You  
did not have to worry that your boss was going to visit porn sites and  
download spyware/adware that would make your software look bad. Etc.

I see very little in the modern design attempts that reflect the old days  
at all.

There's an observation about COBOL to be made here, but I've rambled on  
long enough.<s>



More information about the Squeak-dev mailing list