https://issues.dlang.org/show_bug.cgi?id=15371

--- Comment #1 from bb.temp@gmx.com ---
something like that:

https://github.com/Blumerline/dmd/commit/53419e71b93b0870574d93cbb83ebb7ccbb5e325

or maybe change in the Scope struct the module that indicates where the template should be instantiated (since it function that contains traits code will always be a template).

--

https://issues.dlang.org/show_bug.cgi?id=15371

--- Comment #2 from bb.temp@gmx.com ---
https://github.com/D-Programming-Language/dmd/pull/5289


On n'est jamais mieux servi que par soi-même.

--

https://issues.dlang.org/show_bug.cgi?id=15371

--- Comment #3 from b2.temp@gmx.com ---
Also if a template would be instantiated in the module that use it...

--

https://issues.dlang.org/show_bug.cgi?id=15371

--- Comment #4 from b2.temp@gmx.com ---
In the meantime, when the trait code is for a struct or a class it's possible to use its '.tupleof' property. It's not affected by the visibility.

Instead of all member:

    import std.meta: aliasSeqOf;
    import std.range: iota;
    foreach(i;  aliasSeqOf!(iota(0, T.tupleof.length)))
    {
        alias MT = typeof(T.tupleof[i]);
        ...
    }

https://issues.dlang.org/show_bug.cgi?id=15371

Jacob Carlborg <doob@me.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |doob@me.com

--- Comment #5 from Jacob Carlborg <doob@me.com> ---
A pointer can be used as well to bypass the protection. It works for other symbols than fields.

--

https://issues.dlang.org/show_bug.cgi?id=15371

Martin Nowak <code@dawg.eu> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |code@dawg.eu

--- Comment #6 from Martin Nowak <code@dawg.eu> ---
(In reply to b2.temp from comment #3)
> Also if a template would be instantiated in the module that use it...

Well, there are plenty of reasons to not do that, foremost being that template
instantiations should be independent of instantiation scope.
That's why we have mixin template as an alternative.

--

https://issues.dlang.org/show_bug.cgi?id=15371

--- Comment #7 from b2.temp@gmx.com ---
related https://issues.dlang.org/show_bug.cgi?id=10267

--

https://issues.dlang.org/show_bug.cgi?id=15371

b2.temp@gmx.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |WONTFIX

--- Comment #8 from b2.temp@gmx.com ---
With the direction took by https://github.com/dlang/dmd/pull/6078 this improvement is now impossible.

--

https://issues.dlang.org/show_bug.cgi?id=15371

b2.temp@gmx.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|WONTFIX                     |---

--

https://issues.dlang.org/show_bug.cgi?id=15371

Walter Bright <bugzilla@digitalmars.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           See Also|                            |https://issues.dlang.org/sh
                   |                            |ow_bug.cgi?id=19326

--