On Thursday, 22 September 2016 at 14:03:04 UTC, Andrei Alexandrescu wrote:
On 9/21/16 7:37 PM, Walter Bright wrote:
On 9/21/2016 3:48 PM, Brad Anderson wrote:

Sorry I wasn't clear. The free entry is only for the 8:30 talk.

Slides are nice, I hope the talk was good. His notion of total functions gets a bit weakened by the existence of the default constructor (e.g. if the return type has a default value, you can always write a total function that just returns it). In order to avoid such degenerate cases, he'd need to add the requirement that the function is also injective (maps different inputs to different outputs). Then his examples are meaningful (and beautiful).

In D, the closest we get to the notion of a total function is a nothrow pure function. As far as I know we cannot enforce injectivity.


Just to be clear, this wasn't slides for the talk Walter was attending. I'm not sure if Walter was going to the Lightning Talks, the Concepts/Range talk, or the Biggest Security Fails talk. This was just a separate one from earlier in the week.

I remember you recommending Types and Programming Languages by MIT Press awhile back. Is that still what you'd recommend for learning about type theory?

Reply via email to