From mboxrd@z Thu Jan 1 00:00:00 1970 Subject: Re: [9fans] Interesting in trying out Plan 9 From: Dave Lukes To: 9fans <9fans@cse.psu.edu> In-Reply-To: <716fa16ad5690d5afe47d5ae26b431ec@collyer.net> References: <716fa16ad5690d5afe47d5ae26b431ec@collyer.net> Content-Type: text/plain Message-Id: <1077101443.7471.40.camel@zevon> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Date: Wed, 18 Feb 2004 10:50:43 +0000 Topicbox-Message-UUID: e9f67f2a-eacc-11e9-9e20-41e7f4b1d025 On Wed, 2004-02-18 at 04:06, Geoff Collyer wrote: > Right, at least in Unix processes operated on themselves, sometimes > upon request of other processes. If Linux isn't doing that, they're > in for a world of pain (or locking, at the very least). Tell me about it. We have a bunch of Linux servers here, and every few weeks, some script pukes an irreproducible but subtle error (e.g. bizarre number from some stats analysis). I tear my hair out for hours wondering what's wrong, never find out, chalk it up to gremlins and move on. Now, maybe I've got serious brain damage, maybe it's clever hackers (the confuseSA virus:-), maybe it's cosmic rays, maybe it's an incredibly obscure bug in bash/awk/..., or maybe it's an obscure kernel bug, but the fact is that the whole infrastructure (hideously complicated tools bolted onto an obscenely complex OS) makes any informal (let alone formal) verification impossible. If I could just convince myself that _some_ part of this mountain of manure was _not_ to blame, then I'd be some way to tracking down the culprit, but hacking through bash/perl/kernel is just a WOMBAT. Cheers, Dave.