On Saturday 29 December 2001 08:39 am, you wrote: > On December 29, Collins Richey enlightened our ignorance thusly: > > Others will be able to help you get back to the status quo ante. > > > > Be advised, I've never heard of any good results using the gcc 3... > > compiler. I'm fairly sure that the compiler is indded broken, although > > you may have broken something else as well. Avoid gcc 3.... like the > > plague - you've already experienced its benefits (ie none). > > I've used gcc 3.0.x without incident since it was released in June. > The compiler is not broken. There are some problems with certain apps > and libraries, but this has been the case with gcc for quite some > time -- such as older version of the kernel and glibc relying on bugs > in the compiler. In fact, I've heard reports of success *and* failure > building the kernel with gcc 3.0.x. > > Please defined what "indded [sic] broken" means. There are more > benefits to using gcc 3.0 than you realize -- I've been living with > it for several months and, aside from known misfeatures, it is a better > product than the 2.95 series: new preprocessor, better > optimizations, more standards compliant, and so forth. >
Kurt, thanks for correcting me. I don't pretend to be an expert in these matters, and I respect your superior knowledge, but I keep reading that various products, including the kernel, don't build well with gcc 3. I seem to remember advice from the kernel developers (Linux or ???) not to use gcc 3. I've heard rumblings that kde doesn't wok either, although I've not tried that. I have an older K6/II which doesn't benifit very much from optimizations. When kernel 2.4.17 (-o2 -march=i586, about as plain vanilla as you can get) fails in one of the driver compiles for the isdn series with a message saying the compiler has generated an invalid instruction, I consider it broken. The other possibility is that the kernel is broken; wouldn't be the first time. I'm sure that sometime in the next few months most of the subleties between gcc and glibc and most of the common products will be worked out and that will be the new "standard." In the mean time, it's still too experimental for my tastes. <rant> It would surely be nice if the compiler and library folks could make progress without breaking old things. I still remember (not too fondly) all the havoc that the current glibc generated when it was new. </rant> Thanks, Collins _______________________________________________ Linux-users mailing list Archives, Digests, etc at http://linux.nf/mailman/listinfo/linux-users
