Received: from PACIFIC-CARRIER-ANNEX.MIT.EDU by po8.MIT.EDU (5.61/4.7) id AA08707; Tue, 25 Aug 98 10:15:29 EDT Received: from hq.ljl.com by MIT.EDU with SMTP id AA10128; Tue, 25 Aug 98 10:15:25 EDT Received: from craig.ljl.com by hq.ljl.COM. with smtp id aa17223; Tue, 25 Aug 1998 09:15:35 -0500 Message-Id: <35E2C6C2.AFA56E5E@airnet.net> Date: Tue, 25 Aug 1998 09:14:26 -0500 From: Craig Goodrich Organization: NAIF X-Mailer: Mozilla 4.06 [en] (Win98; I) Mime-Version: 1.0 To: pmitros@MIT.EDU Subject: Standards deviations: Squares without roots ... Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit ... in the Linux ethos. Great summary of the LSA affair on your website. Attached is the editorial I sent to /.; since Rob didn't put it up and I hate to waste my essays, I'm inflicting it on you. Basic message: the whole "standards" issue is phony, like the "crises" regularly announced by government bureaucrats. Whatever has to happen is already happening quietly. Craig (greetings to left-anarchist RMS from right-anarchist me.... http://airnet.net/craig/g4c ) ============================ == "Official Linux Standards" -- Solutions in Search of a Problem == Amid all of the "Standards" brouhaha, many seem to have lost sight of the basic question: What's the problem? Is there a serious need in the Linux community that's not being met? Linus, in his sensible way, answered the question in his recent ZDNet interview: "We have had de-facto standards already," said Linus Torvalds, who developed the operating system. "Now, because of certain commercial interests, certain people seem to think that they need to be written down... it's not a question of changing the operating system per se, it's a question of documentation... writing it down, making it official. And that's going to take time." And he's right. I sympathize with the many users and hackers who have faced endless annoyances getting a package designed for one distribution working in another -- I spent all evening a while back trying to make Star Office 4 work right on a Red Hat 5 system -- but these problems are just that -- annoyances -- and hardly warrant the meat axe of preparing an entire "Standard Base" distribution or forming some grand bigwig consortium of corporate high-rollers. In fact, anyone who studies the situation closely will come to the conclusion that from the point of view of application software, the Linux distributions are remarkably consistent. Mainstream user-oriented binaries like Netscape, Word Perfect, and Applix install without too much hassle on any current distribution. (System-level software is different, of course; that's one reason we *have* distributions, and anyone who tries to install, say, a Slackware init script on a Red Hat system *deserves* whatever problems arise.) ==== Is this an elitist point of view? What about the casual user who doesn't know enough to do even minor tweaking? Admittedly, if Linux is going for World Domination of the Desktop, even those minor glitches between distributions need to be ironed out. Who should do it? A free-software Napoleon who wants to create his own distribution? A free-software Barnum trying to pull rich suckers into his tent? Majority vote of a nerd community that can barely agree on whether the sun will come up in the morning (or whether their Sun will come up this morning)? What is needed, of course, is people intimately acquainted with both the whats and the whys of each distribution, who also have a strong interest in making third-party software as easy to install on their distribution as possible, and who have a pretty good idea of what the future plans for their distribution are. These people need to discuss the situation in an atmosphere divorced from the high noise level and frenzy of both kibitzing advocates and Madison Avenue hucksters, so that the concerns of each distribution -- Debian's views on RPM, RedHat's preoccupation with ease of upgrading, Pat Volkerding's religious reverence for Unix v6, and so on -- can be accomodated. And in fact it looks as though this will be possible with a relatively few simple recommendations, on the order of "both /opt and /usr/local should be available for software installation, though one may be a link to the other" and "both libc5 (minimum patch level xx) and glibc should be available for dynamic linking, at least until libc5 is officially declared dead". This sort of short checklist can be adopted without any great fanfare and without doing violence to the special features of any particular distribution, and most importantly will probably reflect features that were already on the distribution-makers' to-do lists anyway because they are The Right Thing To Do(tm). And this is what's happening right now. RedHat and Debian are looking at differences between their distributions that might cause problems for third-party software; I have little doubt that SuSE, Stampede, and even Slackware will join the discussion once it's clear that these are techies talking about technical solutions to technical problems, rather than some kind of political or commercial grandstanding. As ESR has documented, a consensus about what to do among those who are actually doing it is far superior to any sort of "official" specification. As a side benefit, these discussions will probably also produce some recommendations for third-party developers as to what to depend on (and what to avoid) in their installation routines; this will also be useful to software companies new to Linux -- just, in Linus' phrase, "writing it down." ====== All this uproar over standards has not done Linux any good. Both the LSB and the LSA grossly exaggerated the incompatibilities between distributions for their own aggrandizement, in the same way that some alarmist Y2K consultants use armageddon scenarios to scare their audience of clueless suits out of higher fees. In the process, they may have made some ISV decisionmakers leery of porting their software to Linux. So I suggest we all just shut up about standards, except when some pretentious would-be Standards Czar needs to be stomped, and let the various distributions work it out through the LCS mailing list. There's a lot of coding and advocacy to be done; let's get back to work...... Craig Goodrich Rural Village Systems Elkmont AL