From owner-freebsd-hackers Wed Feb 14 12:06:20 1996 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id MAA09593 for hackers-outgoing; Wed, 14 Feb 1996 12:06:20 -0800 (PST) Received: from slipsrv03.btx.dtag.de (slipsrv03.Btx.DTAG.DE [194.25.2.135]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id MAA09580 for ; Wed, 14 Feb 1996 12:06:10 -0800 (PST) Received: from ermail00.btx.dtag.de ([194.25.2.69]) by slipsrv03.btx.dtag.de with smtp (S3.1.29.1) id ; Wed, 14 Feb 96 20:46 MET Received: by ermail00.btx.dtag.de with (S3.1.29.1) id ; Wed, 14 Feb 96 20:46 MET Message-Id: Date: Wed, 14 Feb 96 20:48 +0100 From: FreeBSD_mail@t-online.de (Martin Heller) X-Sender: 06338993050-0002@t-online.de (Martin Heller) Subject: errors which MUST NOT happen with g++/libg++ To: hackers@freebsd.org Sender: owner-hackers@freebsd.org Precedence: bulk >Bala PERIASAMY (bala@cst.com.au) wrote (in the newsgroup) : >: FreeBSD 2.1.0 >: gcc 2.7.2 (with repo patch) >: GNU assembler version 1.92.3, FreeBSD $Revision: 1.4 $ >: I have got gcc 2.7.2 installed with the repo patch. >: I tried to get libg++2.7.1 install. > You will find that at a minimum that lseek() and > stat() do not work correctly in the lib after > you have fixed the .weak problem. yup ! this is a FreeBSD only problem !!!! lseek works on LINUX (what a shame for FreeBSD ;-) ) > stat() writes beyond the end of its buffer. > lseek() reports EINVAL for good values of the 3rd > parameter. (the same as above) > Not to mention that G++ 2.7.2 SEGV's on a lot of > C++ code that compiles under 2.7.2. on HPUX. > try this class MyException { public: MyException(); virtual ~MyException(); }; extern int dbg_flags; void error(void); void execute(void) { try { } catch( MyException e ) { } if( dbg_flags ) error(); } > BArry this was taken form the newsgroup by me and it shows very clearly the need of an updated gas/ld with '.weak' support . What about asking the guys at CYGNUS ?? Martin