When an exception causes a stack trace to be printed on Linux, only top-level functions are shown. If nested functions are to become idiomatic D style (I use them a ton) then nested functions should be printed as well. Test case: void main() { doIt(); } void doIt() { void nested() { throw new Exception("Test."); } nested(); } Stack trace: object.Exception: Test. ---------------- ./test(void test.doIt()) [0x8057cf9] ./test(_Dmain+0x8) [0x8057ccc] ./test(extern (C) int rt.dmain2.main(int, char**)) [0x805a456] ./test(extern (C) int rt.dmain2.main(int, char**)) [0x805a3b0] ./test(extern (C) int rt.dmain2.main(int, char**)) [0x805a49a] ./test(extern (C) int rt.dmain2.main(int, char**)) [0x805a3b0] ./test(main+0x96) [0x805a356] /lib/tls/libc.so.6(__libc_start_main+0xd3) [0x8c3e93] ./test [0x8057c21]
If I run the program under gdb and set a breakpoint on the throw, gdb is able to display the nested function as part of the call stack. So obviously all necessary info is there, it's just a matter of incorporating it into the stacktrace generated by the throw.
I think the problem here is that previous exception handler was using BPs stored on stack to walk the stack frame, and nested functions were not doing this properly (if I remember correctly, they were not using *BP for the stack frame, or something similar). With the new exception handler, this issue is gone: object.Exception@test.d(8): Test. ---------------- ??:? void test.doIt().nested() [0x422ae9] ??:? void test.doIt() [0x422a92] ??:? _Dmain [0x422a80] ??:? _D2rt6dmain211_d_run_mainUiPPaPUAAaZiZ6runAllMFZ9__lambda1MFZv [0x423126] ??:? void rt.dmain2._d_run_main(int, char**, extern (C) int function(char[][])*).tryExec(scope void delegate()) [0x423070] ??:? void rt.dmain2._d_run_main(int, char**, extern (C) int function(char[][])*).runAll() [0x4230e2] ??:? void rt.dmain2._d_run_main(int, char**, extern (C) int function(char[][])*).tryExec(scope void delegate()) [0x423070] ??:? _d_run_main [0x422fe1] ??:? main [0x422b75] ??:? __libc_start_main [0x935bb82f]