From mboxrd@z Thu Jan 1 00:00:00 1970 From: erik quanstrom To: 9fans@cse.psu.edu, Russ Cox References: Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 In-Reply-To: Subject: Re: [9fans] MS Research reinvents Inferno? Message-Id: <20051214124714.E3B2498B5B@dexter-peak.quanstro.net> Date: Wed, 14 Dec 2005 06:47:14 -0600 Cc: Topicbox-Message-UUID: c58da5b0-ead0-11e9-9d60-3106f5b1d025 this isn't really about the kernel, but i think there are two important bug-reducers in plan9 that windows and *nix systems don't have: 1. plan9 picks a standard and sticks with it. the best example is character set. plan9 picks one and implements it. and things just work. gnu grep doesn't and it has many character-set related bugs. for example, in a utf-8 locale, it's about 100x slower than in a us-ascii locale. 2. plan9's utf-8/unicode implementation also avoids bugs by not implementing bidi, character shaping, combining characters or extended planes. to which most of the code in full unicode implentations is devoted - erik Russ Cox writes | | > It would be interesting to see a bugs-per-million-lines-of-code | > statistic for Plan 9 vs other systems. In my experience, Plan 9's | > advantage is that it's small and simple enough that the bugs are | > relatively easy to find and fix. | | I agree. I started toward figuring that out for Plan 9 by | putting together the kernel history at http://swtch.com/plan9history. | Step 2 is understanding every change and categorizing them, | including whether or not they are bug fixes. Then one can derive | many interesting statistics about the bugs. | | I got about a quarter of the way through categorizing | http://swtch.com/cgi-bin/plan9history.cgi?p=^(pc|port)/.*\.[ch]&v=filelist | I'm hoping to find a student at MIT to finish the job. | | Russ