Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!bloom-beacon!husc6!cmcl2!phri!dasys1!tneff From: tneff@dasys1.UUCP (Tom Neff) Newsgroups: news.admin Subject: Re: A counter-example for those who would eliminate X Message-ID: <5330@dasys1.UUCP> Date: 5 Jul 88 06:12:15 GMT References: <264@octopus.UUCP> <3302@s.cc.purdue.edu> <265@octopus.UUCP> Reply-To: tneff@dasys1.UUCP (Tom Neff) Organization: Independent Users Guild Lines: 55 Pete Holzmann points out that humongous machine-specific *source* postings are appearing on the net recently, and suggests a comparison with PC binaries. He insists that he's not disapproving of, for instance, the Mahjongg tiles, but it seems to me there would be little grounds for comparison if you couldn't make the case that they, like the PC binaries, are signs of a Nasty Trend. And I think he's got a point. If you're going to post source to thousands of machines, most of which are not Suns, then you should have compile-time switches that let the user build versions for other machines. If that's not possible because an application's too hardware-specific, then don't post it to comp.anything -- set up ftp or mail it to requestors. The "source is source" excuse is no excuse. Source can be anything, trash or gold, suitable or unsuitable. Just because something is expressed in source format doesn't give it religious immunity, especially if it needs 18 parts. It had better be good, and general. Unfortunately, none of this gets binaries off the hook. The problem is twofold -- technical and political. Technically you have the difficulty of assembling working binary versions from the uuenc/shar'd/split-up format we use; the difficulty of getting binaries to work on clone hardware and unanticipated system configurations (common problems); and the overwhelming difficulty getting corrections made when bugs are spotted. (You can't do it yourself, the author has to, and authors get awful busy even when they're being real nice!) The political problem is more profound, and has to do with a consensus of what USENET is. There is a powerful and traditional school of thought that says USENET is a medium for stuff humans can read and understand: text articles, tables, poems, source programs and such-like. Binaries aren't readable. There is also a school of thought (the "programmers rule!" school) that says USENET readers ought to be able to *tinker* with anything they grab from the net. You cannot easily tinker with binaries. Now it turns out that programmers, for all their vaunted concern with the source code, are USERS too just like the rest of us, and almost always use software for which they don't have the source or have never read the source. Life is too short (and the employers' personnel dollar too dear) for everyone to compile their text editor before using it, for instance. You just use it. If someone offers you a small, terrific utility that increases your productivity regardless of how the source code looks, you take it. USENET can be as good at this as the next net if it wants; the overall traffic shouldn't be too big. The problem is size. The hugest stuff just doesn't belong. And, of course, some people have 0.00 use for PC anything. That's why I recommend (a) a 64k limit on binary postings (couting all "part" articles) and (b) a separate bin.* hierarchy so that non-PC sites can filter out the binary stuff in a straightforward way. -- Tom Neff UUCP: ...!cmcl2!phri!dasys1!tneff "None of your toys CIS: 76556,2536 MCI: TNEFF will function..." GEnie: TOMNEFF BIX: are you kidding?