ยทยทยท@jpl.nasa.gov (Erann Gat) writes:
> The point is this: the more kinds of errors are flagged for you at
> compile time, the more you can get lulled into a sense of
> complacency and start to believe that if a program compiles without
> errors then it is actually free of bugs.
I tend to think about this differently, that type errors are a set of
measure zero in the space of errors. I haven't tried to formalize
this notion. I wonder what type researchers think about it. Anyway,
that's why I don't believe in static typing. It gives you a little
bit but it takes away a lot.
--
John M. Adams
John M. Adams wrote:
> I tend to think [...] that type errors are a set of
> measure zero in the space of errors. I haven't tried to formalize
> this notion.
Actually, the space of errors is relatively uninteresting, since it has
an countably infinite measure, and we're talking about a finite set of
programs that are written.
What's more interesting is the space of errors that are likely to
happen, each error weighed with its probability. If that space has a
finite measure (and I assume that's the case), it makes sense to talk
about the measure of various error types.
However, I disagree that the type errors have measure zero.
What's lacking here is also a definition of "type error". I have a vague
notion of what it is, but others have shown different ideas (and even
the term "type" doesn't seem to mean the same for everybody).
Which means that the entire topic isn't going to lead to any conclusive
statements. (Some side topics have been interesting though.)
Regards,
Jo