Bug 288 – type of opEquals

Bruno Medeiros schrieb am 2006-08-14: > I don't think this test case is correct. You test the type-of a TypeInfo > (which is an Object) equality operation vs the type-of an int(which is a > primitive type) equality operation. Nothing says that such type-of > should be the same (even though I very much think it should). > > What I do think is not correct is the following (second line) : > > writefln( typeid(typeof(typeid(int) == typeid(char))) );// int > writefln( typeid(typeof(typeid(int) == typeid(int))) );//bool INCORRECT? > writefln( typeid(typeof(new Object == new Object)) ); // int > > writefln( typeid(typeof(typeid(int) != typeid(char))) ); // bool > writefln( typeid(typeof(typeid(int) != typeid(int))) ); // bool > writefln( typeid(typeof(new Object != new Object)) ); // bool http://www.digitalmars.com/d/expression.html#EqualExpression # # Equality expressions compare the two operands for equality (==) or # inequality (!=). The type of the result is bool. # Thus samples 1 and 3 are incorrect. Sources with issues: dmd-0.164/src/dmd/typinf.c:341: * int function(void*,void*) xopEquals; dmd-0.164/src/phobos/std/bitarray.d:289: int opEquals(BitArray a2) dmd-0.164/src/phobos/std/boxer.d:323: bool opEquals(Box other) dmd-0.164/src/phobos/object.d:23: int opEquals(Object o); dmd-0.164/src/phobos/object.d:109: int function(void*,void*) xopEquals; dmd-0.164/src/phobos/internal/object.d:123: int opEquals(Object o) dmd-0.164/src/phobos/internal/object.d:181: int opEquals(Object o) dmd-0.164/src/phobos/internal/object.d:570: int function(void*,void*) xopEquals; Added to DStress as http://dstress.kuehne.cn/run/o/opEquals_05_A.d http://dstress.kuehne.cn/run/o/opEquals_05_B.d http://dstress.kuehne.cn/run/o/opEquals_05_C.d http://dstress.kuehne.cn/run/o/opEquals_05_D.d http://dstress.kuehne.cn/run/o/opEquals_05_E.d http://dstress.kuehne.cn/run/o/opEquals_05_F.d http://dstress.kuehne.cn/run/o/opEquals_05_G.d http://dstress.kuehne.cn/run/o/opEquals_05_H.d

AIUI all expressions involving == and != are supposed to be of type bool. The fact that Object.opEquals is of type int is another matter. See digitalmars.D.bugs:7933, "int opEquals(Object), and other legacy ints". Walter claimed it's "for efficiency reasons", but the cited efficiency reasons don't work here for as long as it's opEquals rather than opNotEquals.

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 [email protected] schrieb am 2006-08-17: > http://d.puremagic.com/issues/show_bug.cgi?id=288 > AIUI all expressions involving == and != are supposed to be of type bool. > > The fact that Object.opEquals is of type int is another matter. http://www.digitalmars.com/d/expression.html#EqualExpression > For class and struct objects, the expression (a == b) is rewritten > as a.opEquals(b), and (a != b) is rewritten as !a.opEquals(b) Thus opEquals should return a boolean result. Thomas -----BEGIN PGP SIGNATURE----- iD8DBQFE5YffLK5blCcjpWoRArZnAJ968NiHssrxp64Laey3wuAUXHKBvwCfeA74 w9FDlhgZht5/ii3rQAqHNN0= =szwS -----END PGP SIGNATURE-----

*** Bug 1989 has been marked as a duplicate of this bug. ***

Fixed in D2.016, D1 is keeping 'int opEquals' for backwards compatibility.