From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 10998 invoked by alias); 30 Mar 2004 23:26:06 -0000 Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Archive: List-Post: List-Help: Sender: gcc-bugs-owner@gcc.gnu.org Received: (qmail 10964 invoked by uid 48); 30 Mar 2004 23:26:03 -0000 Date: Tue, 30 Mar 2004 23:26:00 -0000 Message-ID: <20040330232603.10963.qmail@sources.redhat.com> From: "bangerth at dealii dot org" To: gcc-bugs@gcc.gnu.org In-Reply-To: <20040213062200.14137.mattyt-bugzilla@tpg.com.au> References: <20040213062200.14137.mattyt-bugzilla@tpg.com.au> Reply-To: gcc-bugzilla@gcc.gnu.org Subject: [Bug pch/14137] [pch] ICE in cgraph_finalize_compilation_unit, at cgraphunit.c:407 X-Bugzilla-Reason: CC X-SW-Source: 2004-03/txt/msg03451.txt.bz2 List-Id: ------- Additional Comments From bangerth at dealii dot org 2004-03-30 23:25 ------- In my view, if compilation time is a problem, then a) the project is big, b) we are talking about regular builds, not ones where someone is playing with compiler options, and b) very likely people are using a makefile to build things. It seems very likely that within a Makefile, the number of different combinations of flags is relatively small (say: 2, one for debug, one for optimized builds), and in this case it is reasonable to ask them that they generate one pch file per set of flags. After all, we support this, by putting multiple pch files into one directory. I consider simply requiring the user to use the exact same combination of flags as an absolutely acceptable requirement and am not expecting that we get many bug reports. On the other hand, if we get reports about incorrect code generation due to problems with PCH, this is going to be significantly worse. Please also keep in mind here that PCH bugs are us bugmasters' worst nightmares, since they often come in huge chunks, involve not only a single preprocessed file but a large collection of header files, may involve different compilation flags for different parts of the build, etc. PCH bugs have been very resistant to our efforts in the past, and have most often been analysed by the people who know how PCH is implemented, rather than by bugzilla people. I therefore very strongly urge you to go the safe route and only allow mixing compilation flags for PCH generation and PCH use if we are _absolutely_ (double-checked, verifiably) sure that this is safe! W. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14137