From owner-freebsd-current Sun Dec 17 14:53:44 1995 Return-Path: owner-current Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id OAA00831 for current-outgoing; Sun, 17 Dec 1995 14:53:44 -0800 (PST) Received: from rocky.sri.MT.net (rocky.sri.MT.net [204.182.243.10]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id OAA00823 for ; Sun, 17 Dec 1995 14:53:36 -0800 (PST) Received: (from nate@localhost) by rocky.sri.MT.net (8.6.12/8.6.12) id PAA21983; Sun, 17 Dec 1995 15:56:08 -0700 Date: Sun, 17 Dec 1995 15:56:08 -0700 From: Nate Williams Message-Id: <199512172256.PAA21983@rocky.sri.MT.net> To: current@FreeBSD.org Subject: Impressions of stability ( was Re: FreeBSD-current-stable ??? ) In-Reply-To: References: <7748.819191407@time.cdrom.com> Sender: owner-current@FreeBSD.org Precedence: bulk I just posted something, but I read through the all of the email on this subject and I want to summarize what I'm seeing. S'ren and I are seeing lots of problems with -current. Peter is seeing some problems with -current, most related to the AHC driver. Bruce, John, and Julian aren't seeing any (significant) problems on their systems. So, S'ren and I are upset (understandably), and are venting our frustrations on the mailing lists. The users are seeing this and are un-willing to run -current on their boxes, and are asking for a more 'stable -current' so they can try out new features. Some of the developers are defending the unstability of -current as 'the way things are', which is justified, but this attitude has been going on for a very long time. Me, I resent being told that it's 'operator error', as well I'm sure S'ren does (and probably anyone else). We've been doing this long enough to know what we are doing, and the fact of the matter is that for us the system is unusable. However, this response of 'that's -current' has gone on for a while now. I watch the mailing lists, commit lists, and such and I've yet to see fixes which directly apply to bugs that were reported a long time ago. Now, we could be trying to add so many 'good/tested' code to the system that the bugs we're tracking down will become so obvious that it will be easy to find them, but somehow I doubt this is the case. I agree that the changes made to the tree have been good changes, but what I disagree on is the timing. However, in defense of the -core team and the developers, I can see where an unstable -current is less important since the next release is not based on it. There is going to be alot more time to fix the bugs in -current than normal, so more instability is acceptable. But, the ways things have been going I don't see how -current will ever become a release candidate for 'normal' users unless the 'feeping creaturism' slows, which I suspect will happen in time. In the meantime, *how* can folks help the developers debug -current. Obviously, the system works fine for the developers who added the new fetures, but it's not for others. We need a way for the -current developers to give some time to helping out those folks who can't get it working who *can* help them out. I'm willing to get -current more stable if I can get someone to work with me. I'll have more time to work on it over the holidays as I have a week off, and I will be around home for most of it. Nate