Re: "Near Miss" error handling?

Eric Eide <eeide@cs.utah.edu> wrote:
>
> The INTERLISP DWIM (``Do What I Mean'') facility did something
> similar to this.
> I don't know what became of the local INTERLISP manuals, though,
> so I can't be more specific about how DWIM actually worked.


DWIM tried to correct all types of errors: spelling, syntax (and maybe
even semantics).
It died a none-too-soon death from overambition. (This disease has
killed many AI projects, and DWIM was one of them at its core.) The main
problem was that even if DWIM had caught all errors, it never had quite
enough information to guess what the right correction would have been.
The corrected code was usually worse than the original.


Note that a simple spelling corrector is far from overambitious. Such
mechanisms were quite common in "student compilers" when batch
processing was normal: instead of simply failing the run with a compiler
error because the student mispunched a letter, the compiler would try to
correct the error and continue as far as possible. This has saved me
some of my batch runs, though not as many as I would have liked.


> (I'm not quite old enough to
> have actually *used* INTERLISP, unfortunately!)


I am, son :))


(Fortunately, just one semester of card punching... just enough to
appreciate interactive workstations.)


Regards,
Joachim

Return to the comp.compilers page.
Search the comp.compilers archives again.