Return to BSD News archive
Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!decwrl!nic.hookup.net!swrinde!cs.utexas.edu!howland.reston.ans.net!newsserver.jvnc.net!igor.rutgers.edu!geneva.rutgers.edu!hedrick From: hedrick@geneva.rutgers.edu (Charles Hedrick) Newsgroups: comp.os.386bsd.questions Subject: Re: Shared libraries (was: FreeBSD vs. Linux???) Message-ID: <Feb.5.12.30.49.1994.4735@geneva.rutgers.edu> Date: 5 Feb 94 17:30:49 GMT References: <2igljt$dvc@crl2.crl.com> <2ihgut$2oq@homer.cs.mcgill.ca> <lha.760146931@mullian.ee.mu.oz.au> <2in1j1$cq5@pdq.coe.montana.edu> Organization: Rutgers Univ., New Brunswick, N.J. Lines: 33 Lachlan Leicester Henry ANDREW <lha@ee.mu.OZ.AU> wrote: > >The main advantage I found of 386BSD over Linux was its stability. I >don't have the time or inclination (or possibly the expertise :-) to >recompile the kernel every couple of days. Is there a nice stable >version of *BSD with shard libraries? I've never understood why people consider it a disadvantage that Linus makes his internal testing releases visible. The kernels that change every few days are ALPHA kernels, hiding in Linus' directory in Finland. No one should feel compelled to move to each one, nor should one get the impression that the system is unstable because Linus is continuing to do new work. The folks at Sun are building new kernels just as rapidly -- they just don't let you see them until they're released. The one exception to that is that during the least few weeks, we've been desperately trying to swat network bugs, and have advised some people to try ALPHA kernels. I hope that the current released version will work correctly. So far I haven't seen any reports of trouble with it that didn't sound like configuration problems. (Of course few people give bug reports with enough information that one can really be sure.) However there were enough last-minute changes that I wouldn't be surprised to see a real bug or two. I absolutely agree that unreliability of the network code has been a problem with Linux. Between pl14 and pl15 Linus got personally involved. This has been a great help, as he does a much better job of spotting and fixing race conditions than anyone else. I've also spent a lot of time bringing the TCP layer as close as possible to RFC1122, though I haven't done as much as others such as Alan Cox. I believe that TCP will work reliably by release 1.0, and I think it's very close in pl15. (Obviously it's unsafe to make any stronger statement about pl15 than that until more people have tried it.)