Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!utgpu!water!watnot!watmath!clyde!rutgers!lll-lcc!lll-crg!hoptoad!farren From: farren@hoptoad.UUCP Newsgroups: comp.sys.amiga Subject: Retraction - don't ARC (was Re: Jtime.arc Posting) Message-ID: <1968@hoptoad.uucp> Date: Tue, 7-Apr-87 05:42:40 EST Article-I.D.: hoptoad.1968 Posted: Tue Apr 7 05:42:40 1987 Date-Received: Fri, 10-Apr-87 00:38:56 EST References: <10423@topaz.RUTGERS.EDU> <520@madvax.UUCP> <479@oscvax.UUCP> <1957@hoptoad.uucp> <3047@jade.BERKELEY.EDU> Reply-To: farren@hoptoad.UUCP (Mike Farren) Organization: Nebula Consultants in San Francisco Lines: 25 In a previous article, I defended the practice of ARCing files first, then uuencoding the arc and posting it. Several people, notably John Gilmore and Mike Meyer, claimed that this would be inefficient, since news compresses data as it is sent. Rather than take this statement unsupported (and, because I thought I was right!), I ran a test. I selected nine files, at random, from /usr/local. They were mostly binary, but a couple of text files got in. I then processed them in two ways. First, I ARCed them, and uuencoded the resulting .ARC files. I then compressed the resulting files, to simulate what news does, and recorded the resulting file sizes. Next, I uuencoded them without using ARC, and compressed the uuencoded files. I then compared the file sizes with the original, ARCed/uuencoded/compressed file sizes. I lose. In only two cases was the ARCed/uuencoded/compressed file smaller than the uuencoded/compressed file, and both of those were files less than 2K long in the original. The larger the original file, the worse the discrepancy, with the largest file 25% larger when ARCed than when not. I'm a believer now. Don't ARC binaries. Just uuencode them, and post that. -- ---------------- "... if the church put in half the time on covetousness Mike Farren that it does on lust, this would be a better world ..." hoptoad!farren Garrison Keillor, "Lake Wobegon Days"