Bug 6085 – The filename part of a thrown core.exception.UnicodeException is incomprehensible

Argh, the Unicode seems to make the post disappeared. Let me try again. Test case: ================================== void main() { string s = "\xff\xff\xff\0\0\0"; foreach (dchar c; s) {} } ================================== core.exception.UnicodeException@<hundred lines of garbage skipped> opEqualsMFC6ObjectZb(0): invalid UTF-8 sequence ---------------- 5 x 0x000096d6 onUnicodeError + 66 6 x 0x000151e1 dchar rt.util.utf.decode(const(char[]), ref uint) + 373 7 x 0x00012218 _aApplycd1 + 68 <snip> ---------------- ================================== The problem seems to be that the string __FILE__ sent to onUnicodeError is corrupt. Further more, the __LINE__ displayed is always 0.

The reason of corrupted output is because of a wrong signature. diff --git a/src/rt/util/utf.d b/src/rt/util/utf.d index d7aeac1..cdbc27c 100644 --- a/src/rt/util/utf.d +++ b/src/rt/util/utf.d @@ -28,7 +28,7 @@ module rt.util.utf; -extern (C) void onUnicodeError( string msg, size_t idx ); +extern (C) void onUnicodeError( string msg, size_t idx, string file = __FILE__, size_t line = __LINE__ ); /******************************* * Test if c is a valid UTF-32 character. This patch, however, only gives the __FILE__ and __LINE__ of the druntime function, not the actual user code that emits the error. But I think this can't be fixed without modifying DMD because _aApplycd1 doesn't contain the line information. core.exception.UnicodeException@src/rt/util/utf.d(290): invalid UTF-8 sequence ---------------- 5 x 0x000096b2 onUnicodeError + 66 6 x 0x000151ce dchar rt.util.utf.decode(const(char[]), ref uint) + 390 7 x 0x000121f4 _aApplycd1 + 68 8 x 0x000027e5 _Dmain + 37

druntime pull #20. (https://github.com/D-Programming-Language/druntime/pull/20)