Xref: utzoo comp.lang.c++:2884 gnu.g++:130 Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ukma!xanth!uunet!mcvax!inria!litp!vd From: vd@litp.UUCP (Vincent D'LAC) Newsgroups: comp.lang.c++,gnu.g++ Subject: Re: AAGAGGGGHHHHH Message-ID: <1796@litp.UUCP> Date: 30 Mar 89 18:44:11 GMT References: Reply-To: vd@litp.UUCP (Vincent D'LAC) Distribution: comp,gnu Organization: Lab. d'Info. Theorique et Programmation - LITP - Paris - FRANCE Lines: 42 Dans son article grunwald@flute.cs.uiuc.edu ecrit: . . PLEASE GIVE A WARNING MESSAGE WHEN YOU SEE ``delete foo'' BUT YOU . HAVEN'T SEEN THE CLASS BODY FOR ``foo''!!!!!! . . I just spent 3 days tracking down a bug related to this. A class . definition, the AT&T C++ 1.2 and G++ 1.34.1 compilers both assumed . that a call to __builtin_delete or whatever was just fine. . . Dirk Grunwald . Univ. of Illinois . grunwald@flute.cs.uiuc.edu Well, isn't that the more general big failure of C++ : doing lots of things in your back ? I suggest a little test : just try to enumerate (without the book) the erroneous figures (type errors for example) that you can legally write in this language when you have : a class (say C) with a constructor "C(int z)", a destructor, and a member operator "int()" in it... In all these situations, your program can run perfectly well, until it crashes (or not ). (flames off) I certainly would not program in C++ but i am curious about anything that concerns programming languages. Speaking about C++, I would like to know more about the current practice : 1- Do YOU, C++ programmers have some kind of self-discipline to avoid these kind of traps - any method ? how do you accomodate with such a weakly specified language ? 2- Are such methodologies developped in software companies ? (by the way does any soft.comp use this language for large projects ?). merci d'avance (thanks in advance) Vincent Delacour