Generic actual inference - #243
Conversation
de1e604 to
2aa4834
Compare
| * The `"<"` is deduced as per pre-existing instantiation rules. | ||
|
|
||
|
|
||
| Inference through generic packages |
There was a problem hiding this comment.
This RFC is fundamentally a name-resolution change, so could you specify precisely how inference composes with overload resolution: in what order they run, which entities participate, and exactly which cases are considered ambiguous and rejected.
That precision matters for cost too: implementing this will be significant work in both GNAT and Libadalang, and their resolvers can only do it against well-defined name-resolution rules.
|
In the Prior Art section, you might want to talk about other languages that have done something similar. I asked Google: "What languages support implicit instantiation as part of function call" and here is the response: Programming languages that use generics or templates and perform type inference during a regular function call support implicit instantiation (automatically generating a specialized function for specific types based on the arguments provided). [1, 2] Languages with Implicit Instantiation on Function Calls
|
No description provided.