• d-bugmail puremagic.com (39/39) Apr 08 2014 https://d.puremagic.com/issues/show_bug.cgi?id=12545

d-bugmail puremagic.com writes:

https://d.puremagic.com/issues/show_bug.cgi?id=12545

           Summary: An object with .init breaks std.range.ElementType
           Product: D
           Version: D2
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: minor
          Priority: P2
         Component: Phobos
        AssignedTo: nobody puremagic.com
        ReportedBy: destructionator gmail.com



07:01:11 PDT ---
import std.range;

struct Foo { /* same with class btw */
        void init(int a) { }
}

void main() {
        Foo[] a;
        pragma(msg, ElementType!(typeof(a))); // void(int a)
        auto s = stride(a, 3); // *
}

* std/range.d(2188): Error: variable
std.range.stride!(Foo[]).stride.Result.front.val cannot be declared to be a
function


Since ElementType checks Type.init.front.init, it gets a function type instead
of the element type - void(int) instead of Foo.

Then, stride (among many others) uses ElementType!Range in the function
definitions, we get big errors when the higher order range functions are
declared with them in places.



I'm not sure if this is a bug per se, and is easily worked around by renaming
the function from init to anything else, but it is pretty easy to break and
hard to track down the cause if you don't have an idea of the .init idiom used
in phobos.


I kinda feel that declaring a member called init ought to be disallowed, so
Type.init can be trusted in all code.

-- 
Configure issuemail: https://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------