Date: Wed, 17 Dec 1997 03:05:07 -0500 (EST) From: "John S. Dyson" <toor@dyson.iquest.net> To: imp@village.org (Warner Losh) Cc: dyson@freebsd.org, garbanzo@hooked.net, current@freebsd.org Subject: Re: Pentium optimizations Message-ID: <199712170805.DAA00959@dyson.iquest.net> In-Reply-To: <199712170749.AAA03406@harmony.village.org> from Warner Losh at "Dec 17, 97 00:49:50 am"
next in thread | previous in thread | raw e-mail | index | archive | help
Warner Losh said: > In message <199712170437.XAA01802@dyson.iquest.net> "John S. Dyson" writes: > : We should do better to support the effort, and optionally offer the EGCS > : compiler as a port. My guess is that it won't be ready to be the 3.0-RELEASE > : default compiler. > > the tricky part in doing cc as a port is to make sure that it doesn't > get killed in a make world.... > > i'd love to see some constructive ideas here... > I am really not a toolchain person, so my opinions might be garbage. But, hows about using the gcc moniker for egcs or the non-default gcc? That gcc would call the elf version of ld, and use the elf versions of libraries. If gcc isn't acceptable, some other name like egcc or somesuch might be okay? We would get alot more usage of the (new) compiler if it was in ports, and potentially be able to leverage more support from the gcc/egcs teams themselves. Maybe we aren't into "GPL", but GPLed tools aren't all that bad, and it seems that we should be better supporting GCC and other tools that don't taint the runtime. (The only problem is we are already spread very thin, but frankly, we will get a pretty good multiplier if we can somehow put gcc/egcs into ports.) One thing about egcs in particular, and that is it is a fast moving target, and I have found that it sucks up too much of my time to hack on it (I enjoy it too much), and the port maintainer would also have problems tracking it. (Someone else already mentioned that as a potential problem, I think.) However, getting it into ports and somehow handle the deltas from the EGCS development would be a big win (even from a PR standpoint from the EGCS/GCC community.) I really don't want to see anyone wasting time on trying to somehow mangle EGCS into the ports framework, but if someone who is knowlegeable can do it without lots of trouble (or would have lots of fun doing it), maybe it would be a (very) good idea. -- John | Never try to teach a pig to sing, dyson@freebsd.org | it just makes you look stupid, jdyson@nc.com | and it irritates the pig.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199712170805.DAA00959>