Bug 18569 – std.traits.moduleName problem with enum

How to reproduce the issue, 2 files containing the same code, only the module name changes. aa.d: module aa; import std.traits : moduleName; private enum stuff = 0; pragma(msg, moduleName!stuff); bb.d: module bb; import std.traits : moduleName; private enum stuff = 0; pragma(msg, moduleName!stuff); Command line: dmd aa.d bb.d -main Output of the command line: aa aa Whereas I would expect: aa bb ------------- Replacing: private enum stuff = 0; with: private int stuff = 0; gives a proper result. Replacing: private struct stuff { }; with: private struct stuff { }; also gives a proper result. It does not look like it comes from the "parent" traits, as using: private enum stuff = 0; pragma(msg, __traits(parent, stuff).stringof[7..$]); also gives a proper result.

Actually it is not certain that is a Phobos problem, as replacing the moduleName template within std.traits by that reduced code: template moduleName(alias A) { // __traits(parent) returns "module modName" // so we just retreive "modName" enum moduleName = (__traits(parent, A).stringof)[7 .. $]; } ... leads to a wrong result (ignoring the fact package names are not prepended to the module name in that code of course, which is not the issue here anyway).

Strange issue indeed. HAR on run.dlang.io with different symbol names also gives the same result ! https://run.dlang.io/is/VTCiI5

*** This issue has been marked as a duplicate of issue 14501 ***