I also avoid hand-programming by using a framework that uses an expanded data dictionary to provide code in field and entity triggers, etc. Get it right once, use everywhere.

>>As for programming structures: bad structures feel bad to me. I can feel the collision or the missing parts. That sometimes leads me to recreate that's already known.
>
>Ah, so you do have a representation of these abstract things but it's not visual. Which is actually quite interesting and hard to visualize, even in the sketchy, wireframe way that I do it :).
>
>>A friend (from the FORTH interest group) took a look at what I had created in USCD Pascal and said, "oh, you created a keyed ISAM database." I had never heard of same (in 1984).
>
>And so you did carry your 300kg. We've all been there this way or other - I wrote a sorting routine in turbo pascal to get a report in cobol through, because back then, in 1987, m$ cobol for CP/M had a faulty indexing routine incapable of holding more than 32767 records at once... Hey, that means I did the exact opposite, I had an ISAM but it was faulty, so I turned it into sequential in proper order. Well, well...