On Thursday, 10 July 2014 at 22:53:18 UTC, Walter Bright wrote:
On 7/10/2014 1:57 PM, "Marc Schütz" <schue...@gmx.net>" wrote:
That leaves relatively few cases
Right, and do those cases actually matter?
Besides what I mentioned there is also slicing and ranges (not
only of arrays). These are more likely to be implemented as
templates, though.
I'm a big believe in attribute inference, because explicit
attributes are generally a failure with users.
The average end user probably doesn't need to use explicit
annotations a lot, but they need to be there for library authors.
I don't think it's possible to avoid annotations completely but
still get the same functionality just by inferring them
internally, if that is what you're aiming at...