gabriel rosenkoetter on Fri, 25 May 2001 00:39:29 -0400 |
On Thu, May 24, 2001 at 11:31:33PM -0400, Jason Costomiris wrote: > On Wed, May 23, 2001 at 10:35:26PM -0400, gabriel rosenkoetter wrote: > : Um... that's funny, since binaries built with NetBSD's in-tree > : gcc (egcs-2.91.66) and the optional toolchain gcc (gcc 2.95.3) both > : seem to work just fine together on my NetBSD system. > > Same or different releases of libstdc++? That's the rub. Statically > linking your libstdc++ into the code? That's another way around the > problem. Uh, no. It's because gcc 2.95.3 passes regression tests to the previous gcc releases (including the one that's the stock compiler in all the BSD derivatives, egcs 2.91.66). gcc 2.96 does *not* pass regression tests to the last release of gcc 2.95.3, which is part of the reason that the FSF recommends we not use it. > Let's see, RedHat bought Cygnus. Isn't Cygnus the very same gang of > people that the FSF empowered to take their experimental compiler, > egcs and turn that into the mainstream gcc development version - the way > to gcc 3.0? Swell. Those same people said "don't use gcc 2.96 in release code! It's not ready!" If it were, why didn't they just call it gcc 3.0? (Fwiw, the friends I alluded to who work at "RedHat" actually work with Cygnus, and, though I haven't checked with either lately, I wouldn't expect their any too pleased about the situation.) > You're the one asserting that the RH compiler is more buggy, so that's > on you to compile such a list. I've already provided my example: RedHat feels it necessary to include a separate (older) version of gcc to compile the kernel in their operating system. Name me another Unix-like operating system that has *ever* shipped with separate C (etcetera) compilers for userland and kernel use. The fact that gcc 2.96 does *not* work for everything is sufficient reason not to include it. I don't care what the ideal standards say: the important standard is the software being compiled (namely, the operating system and software that runs on it), and the fact that the shipping gcc couldn't (glad to hear it can now, still not interested in using RedHat ever, thanks) is brain-dead broken. Till Linux is up to the new C++ standard, it's just stupid to use a compiler that enforces that standard. ~ g r @ eclipsed.net ______________________________________________________________________ Philadelphia Linux Users Group - http://www.phillylinux.org Announcements-http://lists.phillylinux.org/mail/listinfo/plug-announce General Discussion - http://lists.phillylinux.org/mail/listinfo/plug
|
|