Re: choice of character for relational division

David Cressey wrote:
> ...
> MDL, a variant of Lisp (1971) did somewhat the same thing. "Real MDL code"
> was not the text version of source code as such. Instead, it was the data
> structure that the reader (which contained the parser) turned it into,
> "object code" if you like. The printer had the inverse capability, which I
> guess you could call de-compiling.
>
> This had the advantage of allowing machine generated code to be decompiled.
> If you looked at this stuff with a text editor, you weren't looking at what
> a human orginally wrote.
> ...

Another advantage was that ordinary users saw a canonical form when the object was de-compiled. I had several fractional de-compilers which I found useful for getting the big picture fast.

Also, there was never any question as to which version of source was in play and nobody ever lost source code. (I remember wasting two weeks with an anxious customer figuring out why an HP port wouldn't work only to discover that QC people, in a hurry, had broken the rules and changed some headers.)

p Received on Mon Apr 02 2007 - 15:05:21 CEST