From owner-freebsd-current Sun Aug 11 16:45:09 1996 Return-Path: owner-current Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id QAA19629 for current-outgoing; Sun, 11 Aug 1996 16:45:09 -0700 (PDT) Received: from rover.village.org (rover.village.org [204.144.255.49]) by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id QAA19624 for ; Sun, 11 Aug 1996 16:45:06 -0700 (PDT) Received: from rover.village.org (localhost [127.0.0.1]) by rover.village.org (8.7.5/8.6.6) with ESMTP id RAA25908; Sun, 11 Aug 1996 17:45:01 -0600 (MDT) Message-Id: <199608112345.RAA25908@rover.village.org> To: Josh MacDonald Subject: Re: Whither gcc 2.7? Cc: current@freebsd.org In-reply-to: Your message of Sun, 11 Aug 1996 13:37:26 PDT Date: Sun, 11 Aug 1996 17:45:00 -0600 From: Warner Losh Sender: owner-current@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Josh, Thank you for your comments. : Recent mail on the gcc mailing lists suggest that the -V and -b : options to allow a single driver to use multiple installations : doesn't really work. See subjects heading '-B or -V vs __GNUC__MINOR__'. Other mail in that thread suggests using -B doesn't work, because -B isn't the correct way to select versions or machine types. It also suggests that -V or -b do work correctly. I had to use -V back in the 2.3ish and 2.4ish days, and it worked then. I've used the -b option many times in cross compiling experimental kernels for my MIPS PC with 2.5.x, 2.6.x and most recently 2.7.2. Some quick tests here shows that -V seems to select the right binaries, and define __GNUC__ and __GNUC_MINOR__ correctly. If something has broken between what I'm running and the latest release, then it should be easy to fix. : I think the most important thing is that the gcc port is somewhat : official and blessed by someone who really knows FreeBSD and gcc. : : I found it disturbing that in order to do c++ development on FreeBSD : over the last 9 months I've had to do my own ports of gcc several : times. It gives me an uneasy feeling that I haven't gotten everything : right. I don't care which compiler compiles my kernel or my standard : installation. I've done the port enough times to know that I've not caught everything, but will be happy to accept any critisisms from such a port from those who know some of the details better than I. I can think of several such people on this list (but hesitate to list them all because I'd no doubt slight someone by leaving them off the list by mistake). It is my perception that they are busy with other things and do not have the time to put together a full, bmaked release right now. I have a prototype of the port done, and gcc is building now (with my mere 486 that is overloaded with other things too, this takes a while). I will no doubt have to tweak it here or there before it is ready. I'm torn between having it do a full make bootstrap each time, or just having it to do the normal make. make boostrap is much safer (especially if you don't know what version of gcc you are starting with), but take quite a bit longer to build because it effectively builds three times (once with the native compiler, once with the new compiler compiled with the native compiler and once with the new compiler compiled with the new compiler). Comments? Warner