Path: utzoo!utgpu!watmath!att!ucbvax!tut.cis.ohio-state.edu!brutus.cs.uiuc.edu!flute.cs.uiuc.edu!grunwald From: grunwald@flute.cs.uiuc.edu (Dirk Grunwald) Newsgroups: comp.lang.c++ Subject: Re: service for gnu software (was: Re: C++ 2.0 pricing and licensing policy) Message-ID: Date: 28 Jul 89 19:49:13 GMT References: <1481@ns.network.com> <8723@thorin.cs.unc.edu> <6590188@hplsla.HP.COM> <5670@pt.cs.cmu.edu> <8842@june.cs.washington.edu> Sender: news@brutus.cs.uiuc.edu Reply-To: grunwald@flute.cs.uiuc.edu Distribution: comp Organization: University of Illinois, Urbana-Champaign Lines: 25 In-reply-to: peterd@june.cs.washington.edu's message of 28 Jul 89 18:31:45 GMT re: bugs in AT&T C++ vs. bugs in G++ I've used both (cfront 1.2.1, g++ various versions). Except for a brief period of bugs, G++ has consistently been less buggy. Furthermore, updates are frequent and bugs *do* get fixed. Our version of cfront still doesn't complain if you delete an instance of a class you haven't seen the definition for (bad hole: what if the class defined delete?) + Try using more than 9 arguments to ``form'' in cfront. + Try having a class contain an instance of a class using a destructor. + Try using inlines for non-trivial functions, like spinlocks + Try getting cfront to accept a gnu format ``asm'' so you can try to do spinlocks inline. ....and then, try to get a `cfront' program debugged. The fusion of GDB and G++ make it significantly better than cfront. -- Dirk Grunwald -- Univ. of Illinois (grunwald@flute.cs.uiuc.edu)