DIP31: compile time paradoxes defined. Requested examples added

[deadalnix via Digitalmars-d]

On Thursday, 25 September 2014 at 16:49:27 UTC, Timon Gehr wrote:
> On 09/25/2014 06:53 AM, deadalnix wrote:
>> http://wiki.dlang.org/DIP31
>>
>> I've been asked to add examples of resolution for various 
>> cases (hint
>> they all error out).
>
> "This DIP define their behavior properly, "
>
> Not really. For one thing, at the moment, the DIP doesn't 
> really make clear when and how a 'dependency' occurs.
>

This is why I'm working on a DIP for identifier resolution. 
Indeed, this DIP depend on identifier being resolved in a precise 
manner to get predictible results.

> "Mixin are resolved in order, which mean the first mixin is 
> resolved first."
>
> What do you mean? There is no order; there is no 'first' mixin.

In order in the file. This is already specified in the DIP. Yes, 
when several modules are involved, there is an hole. I have no 
idea how to plug that one.