The most recent version of these essays can be found at http://linuxmafia.com/~rick/faq/.

Rick's Rants

("That's what you get for swimming in the shallow end of the gene pool.")

-Virus-    -Linux Tire-Kicking-    -Proprietary Warez-    -Hardware-    -Netiquette-    -Crybaby-    -Modems-    -MacLinux-    -Miscellany-   

Economy of expression is a good thing. So, rather than have to repeat myself continually, I'm posting my top rants here, for ready reference. Many of you (readers) will be visiting today because I pointedly referred you to the "#"-tagged URL of some particular item, below.

Linux Tire-Kicking Department

• "I'd like to try Linux. Where can I get Linux at minimum cost? Aren't distribution CDs/DVDs available for US $2.50? (Isn't it supposed to be free?)

  You (someone interested in "trying Linux") should not try Linux at minimum cost.

  US $2.50 single-CD/DVD sets are available from many sources, including www.linuxcentral.com (Linux Central) and www.cheapbytes.com (Cheapbytes / Linux System Labs), but these are best understood as a convenience for experienced Linux installers, since what you get is a single binary-only CD/DVD (no source code) in a paper sleeve with no printed documentation, no technical support, and no installation floppies (which you must then, if you need them, generate using a DOS utility from files on the CD/DVD). By the way, it's never just $2.50, since you tend to incur an exorbitant shipping fee plus sales tax (where applicable).

  The point is that, if you are installing Linux for the first time, you already face enough challenges without having no printed documentation, doing without support, and being obliged to construct your own boot floppies.

  The above-cited companies also sometimes sell bundles of the basic CD/DVD plus good tutorial books such as Running Linux by Matt Welsh, Lar Kaufman, Terry Dawson, and Matthias Kalle Dalheimer, 4th edition, from O'Reilly Associates. That would be an improvement.

  Far better for beginners, I urge you to consider boxed sets with printed documentation and paid technical support from such distribution-creators as SUSE, and MandrivaLinux. These tend to run about US $40 each, and are available in stores or directly from the companies.

  I urge caution about Linux distribution CDs/DVDs included inside books: They are almost always several revisions out of date, and tend to have the same disadvantages to beginners cited for Cheapbytes's/LSL's/MadTux's $2.50 offerings.

  (To be fair to MadTux, that CD/DVD vendor is rare and impressive in offering technical support to its customers. Bravo!)

  Linux is "free" in the sense that it (the kernel with bundled hardware drivers) is freely usable for any purpose other than turning it into a proprietary product. That does not prevent anyone (including you) from selling it for any price you can get, with or without extra merchandise such as additional software, books, floppies, technical support, et al.

  As the saying goes, Linux is inherently free in the "free speech" sense, but not the "free beer" sense.

  (One increasingly popular option is that of "live CD" Linux distributions such as Knoppix, which let you try out Linux without touching your hard drive.)

  In any event, I would always recommend buying and reading a good tutorial book such as Running Linux by Matt Welsh, Lar Kaufman, Terry Dawson, and Matthias Kalle Dalheimer, 4th edition, O'Reilly Associates. You should also consider a good reference book, such as Linux in a Nutshell by Ellen Siever et al., 3rd edition, O'Reilly Associates. Therefore, you should budget at least US $100 for your distribution plus needed books.

• Can't I just download Linux?

  Wrong question. You meant to say "Should I download Linux?" If you have to ask, the answer is almost certainly no: A typical Linux distribution consists of some 600MB or more of compressed files. If you were to retrieve them all, would you have somewhere useful to put them? Would you be able to install from that location?

  Your answers to those questions might be "yes", in which case by all means do it. Most people will find it cheaper, faster, and less complex to use more conventional methods.

• Which Linux distribution should I get? Which distribution is friendliest to new users? Should I get Ubuntu?

  Linux poses three distinct challenges: building, administering, and using the system. You might be ecstatic with a Linux system constructed and configured for you, but will (if you're an "ease of use" person) probably be unhappy with the unfamiliar challenge of loading any operating system on Intel-type PCs. (MS-Windows is typically mis-perceived as "easy to install" by those who never install OSes, and who use whatever comes pre-loaded.)

  You can buy pre-assembled, pre-configured Linux systems from many vendors. Those I know of in the San Francisco Bay Area are included in my Other Local Linux Resources list. Please note that all of them do mail-order business.

  The questions of which distribution is "best" and which is "friendliest" are both inherently debatable: Most opinions you'll hear will be both bigoted and based on incomplete, out-of-date information concerning most (or, often, nearly all) alternatives.

  Anyone who tries to give you an easy answer to either question is trying to sell you something.

  You, for your part, should think long and carefully before you ask such questions: Are you even serious about trying Linux at all? How are you going to distinguish between competent, relevant answers and blasts of hot air from people barely less ill-informed than you are?

  No, you should not automatically gravitate towards Ubuntu, just because that distribution is best-known in the USA. (Before 2011, this FAQ item used to say "you should not automatically gravitate towards Red Hat" rather than "Ubuntu", but the distribution relentlessly touted to the masses changes from year to year.) One of the glories of Linux is the richness of choices that you can sample many of at low cost. Consider trying several of them consecutively.

  I personally strongly prefer the Devuan (and formerly Debian) distribution, especially for servers. However, newcomers should consider starting with Bodhi Linux, Linux Mint, Ultimate Edition, MEPIS Linux, or PCLinuxOS, for desktop Linux machines (not Devuan/Debian).

  The (relentlessly marketed) "Ubuntu" distribution family, comprising Ubuntu Linux / Kubuntu / Lubuntu / Xubuntu (a single base distribution with your choice of four desktop environments) is focussed on new users but has ongoing moderately serious bug problems (less so in the "LTS" = Long-Term Support releases), omits support for proprietary AV formats (including MP3s) and Hollywood-type movie DVD playback for understandable reasons of legal complications, omits some similar popular proprietary software such as proprietary Sun Oracle Java, and popular proprietary hardware drivers such as those for Nvidia and ATI video chips (albeit quite good open source drivers are provided), for some problematic wireless chips, etc.

  Linux Mint and Ultimate Edition, cited earlier, are Ubuntu-family distributions from outside maintainers that merge in those "desktop enhancements" so they work out of the box, but are otherwise pretty much the same as standard Ubuntu. Bodhi Linux is also an Ubuntu variant, using a fork of the cutting-edge but relatively light Enlightenment E17 desktop environment.

  If you want a cutting-edge but at least somewhat new-user-friendly Linux distribution but are not fixated on out-of-the-box support for proprietary AV formats, Hollywood-type movie DVD playback, etc., then OpenSUSE, Kademar, Siduction, and Mageia all merit consideration.

  If you're undecided on the question, read Karsten's Distributions Guide, and consult DistroWatch, first. (Warning: Karsten's pages seem to have been unmaintained since 2006, so their specifics are outdated, but the general principles he teaches are sound.)

  You may also find it interesting to compare the "look" of different desktop environments and window managers at the "Window Managers for X" pages.

• Can I install Linux on my { 386sx | 386 | 486 | Pentium | K6 | PPro | PII | m68k | SPARC3 }?

  Every few years, I have to revise this item, to restore realism: This is the 2019 revision.

  The fact is, most computer hardware doesn't age well, especially motherboards and CPUs used for general-purpose computing. At this date, anything older than an Intel Core 2 (2006-1011) or AMD Phenom (2007-2008) is too slow, too fragile from accumulated wear, too expensive / impractical to get parts for if/when parts die, and insufficiently expandable without unjustifiable expense better applied to something modern. (In general, I suggest ten years as a rough cutoff age.)

  One can still justify using machines as old as Pentium III boxes for dedicated server roles, e.g., being just a CD/DVD-burner, just an SMTP mail server, just a DNS nameserver, just a file/print server, just a router/Internet gateway, just a dial-in terminal server, just a data-collection system, just a network monitor, just a firewall, or just an X graphics terminal.

  As a Linux novice, you won't be setting these up right away, but may want to remember the possibility for later on.

  For general-purpose workstation/laptop machines, total RAM is usually the limiting factor: Currently, anything with under 4 GB RAM is sub-par, and will be either disappointing, a specialty project requiring expertise at post-installation configuration, or both.

  Similarly, PowerPC machines such as pre-Intel Macintoshes are all specialty projects at this point, not to mention being really old.

• I want to try Linux, but keep my existing Windows setup (and keep all its files intact). What's the easiest way to do this? Should I use Partition Magic?

  Partition Magic is the best known of several utilities that promise the ability to safely shrink your MS-Windows partition, thereby creating unallocated disk space that you can later re-allocate to Linux-native partitions (which are organised differently, and not comprehensible to MS-Windows). There are also many competitors.

  (This page used to include a long section inventorying & describing proprietary and open-source alternatives, but happily that has long been essentially unnecessary: The functionality to non-destructively split and resize both MS-Windows and native-Linux partitions is now so well integrated into all Linux distributions that you just boot one of the latter and it's just there and semi-automated.)

  Adjusting your MS-Windows setup to clear room for Linux, without harming anything, sounds like just what you need, doesn't it? Unfortunately, it's not that simple. Pull up a chair and get a fresh cup of coffee; it's a long story.

  Typically, your story goes something like this: You have just one computer (and no network other than Internet access), which came pre-loaded with MS-Windows. The machine has a single hard drive, which is organised as a single "drive letter" (partition), which you think of as "C:". (Maybe, alternatively, you have one or two other partitions, but there's no unallocated hard disk space.) You're probably not very familiar with the partitioning process, or that of installing MS-Windows on empty drives.

  Some of your programs came with the system. You bought a few others, got some from friends, maybe some on removable media or from somewhere on-line. All the work you've done with the operating system and software — program and desktop configurations, bookmarks, address books, all sorts of data files — is on there somewhere, but you really don't know how to separate them out and make safety copies of the data elsewhere. For that matter, if you had to reinstall your programs (or those plus MS-Windows), you're not sure you have everything you need, or know how. For example, your PC vendor may not have included a real OS installation CD/DVD, leaving you helpless to ever reload.

  Backups? You know that, in theory, backups are a good idea, but you don't have backups of your MS-Windows machine's contents, and lack the means (e.g., no tape drive). So, you don't think about it much, but try to be careful and hope for the best.

  However, the question of Linux partitions has now come up. You know you don't want to lose anything, and are vaguely aware that you've been flying without a parachute. You're seeking reassurance, and Partition Magic et al. seem to offer it.

  Guess what? They don't. Every such program, in its fine print, warns you to secure a good (tested) backup before using it, and disclaims responsibility if you don't. Resizing is very unlikely to fail, but it could. You could lose everything.

  What I'm suggesting is that, if the possibility of Partition Magic failing makes you anxious, then you have bigger and more fundamental problems that you need to fix first. You should start ensuring that you save all data files to distinguishable, non-program directories (folders). Start finding all your program installation disks and licence "key" (or activation) codes. Find MS-Windows driver software for all of your PC hardware, including a bootable DOS floppy with drivers for your optical drive. Ensure, in short, that you have everything required to reinstall your hard drive's contents from scratch, if need be.

  Why? Because, even if you forget all about Linux and the repartitioning problem, you're still at risk. Hard drive failure, accidental deletion, or viruses can clobber your files, and MS-Windows degrades over time with normal usage patterns anyhow, requiring reinstallation from scratch typically every 6 months to a year, to keep the system running well. Add a Linux dual-boot configuration, and you simply add more risks to an already bad situation.

  If you decide to play the percentages and use Partition Magic, etc. without known-good backups (plus the knowledge required to use them, etc.), you are strongly advised to first use a disk repair utility to do a health-check/repair on your drive volumes, and then defragment your partitions. Be aware that files physically towards the centre of the drive with "read-only", "system", or "hidden" attributes may prevent the defragmenter utility from working. (Such files must be individually found and dealt with.)

  If you're slightly more cautious, and secure tested backups before proceeding (as you were warned to do), please consider: If your backups are reliable, why not rely on them? That is, why not delete your existing MS-Windows partitions, create smaller, empty ones that leave some space unallocated, then reload MS-Windows and restore your backup contents? The advantage is that you'll now have fresh partitions created by the most-appropriate tool, rather than well-worn ones worked over by a third-party tool (e.g., Partition Magic). In the future, you can gain even more advantage by backing up only data files (and program-settings files) and reinstalling your programs from their installation media, which gives you effectively a complete system rebuild.

  I've heard some MS-Windows users object that they have no confidence in their backup programs. Well, then, that's what you should fix, isn't it? E.g., if you suspect the MS-Windows Registry won't be captured correctly, try exporting it to a text file in RegEdit before doing your backups. Then, if necessary, it can be re-imported via RegEdit, later.)

  You will be rare if you take any such pains, but you will also be rare in not suffering massive data losses at random intervals, as most MS-Windows users do.

  While you're pondering these long-term issues, please also consider alternative ways to try Linux: Companies are now throwing away old Pentium III and Pentium 4 machines without a second thought. Pick one up as a second computer, and (given at least 1GB of system RAM) it could make a very reasonable Linux desktop box , even by the standards of your high-end MS-Windows PC. (Save the expensive horsepower for your grossly inefficient Microsoft OS, which needs it.)

  Your MS-Windows computer will benefit, because you won't have to risk its delicate health with repartitioning and byzantine dual-boot setups. Additionally, you can cheaply network the two machines and use Linux the way it was intended, as a networked operating system running long-duration processes (which isn't possible with dual-boot). Install VNC Server on the MS-Windows box, run it "headless" by moving the expensive monitor to your cheap Linux box, and enjoy being able to run both environments from the same comfortable Linux desktop. Let Linux handle your Internet connections for both boxes, provide network services such as Samba to both, handle your backups, etc.

  Having two machines, each playing to its OS's strengths, is better than running one overburdened box with a split personality, isn't it? Forget about dual boot; it entails setup and operational headaches, and provides zero synergy.

• Why isn't Linux doing more to help "Joe User"? Average users don't want to have to become computer experts just to get their work done. Linux installers mustn't demand users make decisions about partitioning, package selection, boot configuration, startup services, and the like. They need a uniform, simple, graphical desktop system with familiar office-productivity applications. Linux needs to accommodate such average users.

  There seems to be a persistent delusion that if you post the above sort of stuff in random on-line media often enough, some specific Linux distribution (or all Linux distributions) will spontaneously mold itself into whatever the speaker wants, and plop itself on his doorstep, wagging its tail. However, the world does not work like that.

  By and large, what happens with and to Linux is entirely in the hands of the Linux community, the core of which are programmers spread all over the world. Almost all of those programmers are volunteers, concentrating on what problems they consider important. If you try to tell members of the Linux community what they "must" do, most will smile, mentally tick you off as some who doesn't get it, and ignore you. A few will take the time to gently explain to you where and why you have misunderstood the fundamentals.

  It is a common misconception that legacy proprietary OSes (such as Mac OS and MS-WinXX) are "simple". They are actually quite complex, but have been pervasively idiot-proofed: Almost all of their complexity is sealed off from the user. Linux and its BSD brethren are at least as complex, but with the full complexity user-accessible, because the user community (overwhelmingly) likes it that way.

  Unlike the case of typical legacy proprietary OSes, Linux can be (and is) almost infinitely reshaped, reconfigured, and repackaged for different roles and situations. That's part of why there are many diverse Linux distributions. And you (or anybody) have an absolute right to create, duplicate, and sell any type of Linux-based system you want.

  So, if you preach to the Linux community about what they "must do", they'll just wonder why you don't do it yourself, or pay someone to, if it's so important to you.

  The Linux-based system you want is almost certainly out there, if only because you can pay someone to set it up exactly the way you want it — and it'll stay that way, once configured, because it's a stable system with meaningful security.

  And that system might be a shrink-wrapped distribution out of the box, or it might be somebody's configuration of one. It might cost money, or it might be free. But, one thing's for sure: If you demand that a randomly-selected member of the Linux community find or point it out for you, he'll probably tell you to get stuffed. With good reason.

  Many people who ask this question essentially mean that they find confusing Linux's ability to be many things for many people simultaneously (even the various users of a single Linux box can have wildly different desktop environments), and to do a dizzying array of both complex and simple tasks — and want these abilities to be removed. Tough. They're there, they're fundamental to the system, and they're gathering momentum. Linux is exploding across the network and into all sorts of machine roles — not just the single-user graphical-workstation role that's the only thing "average users" can think of when they envision computers.

  That's not to say that you can't hide built-in capabilities you don't want to deal with. That's part of what configuration (and choice of Linux distribution) is all about. But they'll still be lurking in the background, and most Linux users will be concerned with the more-complex (and interesting) aspects. If that bothers you, Linux may not be for you.

  Of course, many otherpeople asking this question really mean that they're used to legacy proprietary systems, want to stick with them, and want to feel good about ignoring Linux. If that's the case, relax and stick with what you like. Ignore Linux as long as you like. We don't mind; we're not out to convert you. We're just creating a fabulous, richly detailed software world that is already thriving. Join us when and if you wish.

• Are there any IDEs (Integrated Development Environments) for Linux?

  Sure. A big long list of them, in fact.

  In the longer term, you should consider the possibility that you're solving the wrong problem: That is, long-time Unix developers find that emacs's (and xemacs's) or jed's tight integration with gcc/g++, ctags, and gdb makes them more productive than do the RAD / GUI-builder / IDE tools listed at the above link. This is clearly a question of taste, but please consider the traditionalist Unix user's toolkit, as well.

  You should also consider whether that the language and libraries one picks aren't considerably more significant than are glossy editing tools.

  (This "rant" item dates from the 1990s, when it was common to troll Linux users with claims that no IDEs exist for Linux, which then drew out emacs/gdb fanboys who fell for the bait and verbosely sermonised about why IDEs are Bad and Wrong, with the trollers then claiming victory and repeating the cycle a month later — when in fact the fundamental assumption about Linux IDEs was massively wrong. So, I posted the roster to short-circuit the above-described tiresome rhetoric.)

• What's a good Linux Web browser?

  Short version: I recommend both LibreWolf and ungoogled-chromium as very modern, fully open source graphical Web browsers that respect user privacy and autonomy — based on the Gecko and Blink rendering engines, respectively (and whose evil twins are Firefox and Google Chrome).

  On the horizon (as of 2024), look forward to the release of Ladybird as a new, modern, open source browser (based on its LibWeb browser engine). Also, Servo, an innovative, modern browser rendering engine (not yet a full browser) coded in the Rust language, is in late public beta and worth trying.

  By the way, any graphical Web browser should be checked at coveryourtracks.eff.org to see how well it resists tracking and fingerprinting. If yours gets a lacklustre rating, you should find out why and fix that.

  A case can be made for several forks of 2017's Firefox v. 52 (collectively, "XUL-type browsers"), that fork of Firefox v. 52 Gecko engine being now dubbed the "Goanna" engine (at least, Palemoon calls its hard fork of early Gecko that), that preserve its support for old-style XUL/XPCOM extensions and (accordingly) user configurability, avoid the reduced functionality of WebExtensions that replaced XUL/XPCOM, and are not under Mozilla, Inc. control: Pale Moon (1, 2), Basilisk, Waterfox, Iceweasel-UXP (1, 2), and Iceape-UXP (1, 2). Likewise, some like SeaMonkey (formerly Mozilla Application Suite, formerly Netscape Communicator).

  Joe Brockmeier reviewed Firefox forks on March 4, 2025, in Linux Weekly News.

  Others: Conkeror (extensive keybinding support, in 2024 best run atop Pale Moon or Waterfox), Dillo (lightweight), Luakit (1, 2 —lightweight), surf+tabbed (lightweight), qutebrowser (lightweight, keyboard-focussed), Arachne, GNOME Web (1, 2 — development codename Epiphany), Dooble (1, 2), ADMBrowser (for kiosk use), Falkon (1, 2 —lightweight, formerly QupZilla), chromium (but ungoogled-chromium better respects your privacy), Gnuzilla and GNU IceCat (1, 2), Midori (1, 2 — relatively lightweight, clean UI), NetSurf (small, fast), Liri Browser (fast, clean design), Min (1, 2 — small, fast), Konqueror (1, 2), Otter Browser (1, 2 — inspired by the Opera browser, based on Qt5), Brave Browser (1, 2) all have their fans. (I have some problems with Brave Browser's design choices, but concede merits.)

  Why not Firefox? Mozilla Firefox (formerly Firebird, formerly Phoenix), which has ESR (Extended Support Release) and Release builds, Firefox Beta, Developer, and Nightly (nightly builds; formerly Firefox Aurora, formerly Mozilla Minefield) is, yes, open source, but there are problems.

  Mid-2016 brought the first of two apocalyptic changes to Firefox. First, starting with Firefox 48 on 2016-08-02, Firefox (Release and Beta) refuses to run any extension not cryptographically signed by Mozilla, Inc. at AMO. A temporary workaround of setting preference "xpinstall.signatures.required" to "False" in about:config no longer works in Release and Beta builds. It may work in sundry Developer Edition, nightly, and unbranded builds.

  Most observers appear to have been unbothered by this change (and Mozilla, Inc. produced a list of justifications), but I'm troubled: An open-source application where I cannot decide for myself what ancillary code I'm permitted to run, but need sign-off by a corporation operating the codebase as a walled garden, isn't really open source. Code-signing is fine and laudable — provided I am allowed to have my own signing key in the approval keyring, and Mozilla, Inc. doesn't allow that.

  And what has Mozilla, Inc. done with this control? One, they've been occasionally guilty of installing extensions without user permission. Two, in 2019, they announced they would remotely and retroactively disable extensions in users' browsers they deem insecure. Three, in 2019, they flubbed renewing an intermediate signing certificate on Mozilla servers and thereby briefly locked and disabled all Firefox extensions. Four, starting 2022, they region-blocked Firefox users in China from installing effective privacy extension uBlock Origin, reportedly because Chinese adtech/video vendors had sued Mozilla, Inc. Five, in June 2024, they region-blocked Firefox users in Russia from installing extensions usable to circumvent Russian Federation censorship (at Russia's request), reverting that move two days later, following outcry.

  Six, in Febrary 26, 2025, Mozilla rolled out astonishing Terms of Service and revised Privacy Policy documents that gave the corporation the right to surveil user data for LLM-training or datamining for targeted advertising and related purposes. After some days of overwhelming rejection, Mozilla, Inc. partially retreated, but the loss of faith in the company's good will seems permanent.

  And that, reader, illustrates why you should never permit a corporation to decide which software you're allowed to install and run. (This FAQ item used to chronicle many steps necessary to, also, curb Firefox's many ways in which it "phones home" to Mozilla, Inc. But now you can just run LibreWolf or one of the XUL-type Firefox forks, and not need to.)

  The apocalypse that got many people's backs up was Firefox dropping, starting with Firefox 57 (aka Firefox Quantum) on 2017-11-14, the interface for XUL extensions and related XPCOM object framework, replacing them with the greatly less capable WebExtensions API. Mozilla, Inc. articulated its reasons (there being advantages).

  It is troubling that WebExtensions extensions can be used for user-fingerprinting, destroying privacy. This is doubtless part of why LibreWolf recommends minimal browser add-ons: "Every extra piece makes your attack surface larger, and is likely going to make you stand out".

  Useful WebExtensions-type extensions (for modern Gecko engine browsers): DontBugMe, AdBlockPlus, NoScript, and RequestPolicy Continued. High recommendation for Luminous, which elegantly returns to the user control over Javascript events, and among other things stops sites from preventing user access to Clipboard. Worth checking: Disconnect Private Browsing & Disconnect Private Search, Self-Destructing Cookies, and Tridactyl.

  Useful extensions for XUL-type (Goanna-like engine based) browsers: User Agent Switcher, possible alternatives to vanished BugMeNot, Firebug, OptimizeGoogle but that unfortunately was orphaned in 2012 so works less well, RequestPolicy, and Beef Taco. Worth checking: Clean Links, and CertWatch or Convergence (unmaintained since 2012) or Monkeysphere (or other, similar ideas).

  Potentially useful is the GNU Project's list of XUL-type extensions and Martin Brinkman's Ghacks list of privacy/security tweaks.

  A few suggested extensions for Google's Chromium browser (or, better, the ungoogled-chromium variant): ScriptSafe, Alternate Tab Order, uBlock Origin, uMatrix, and Tampermonkey.

  However, the drawback of relying on Chromium or any browser based on Google's "Blink" rendering engine in 2024, when Google announced that Chromium would abandon (1, 2) the "Manifest v2" extension interface for Chrome-family browsers, with the obvious intent to prevent ad-blocking extensions such as uBlock Origin from functioning, any longer.

  Some users like the proprietary Opera, Vivaldi, Slimjet, Floorp, SRWare Iron, and Google Chrome (which is a surveillance engine: 1, 2, 3) browsers. I disrecommend them.

  (Above are graphical Web browsers. For completeness, these text-console Web browsers are also around: ELinks, links (text mode), lynx, w3m, emacs-w3m, w3mmee, Netrik, Retawq, Line Mode Browser, Debris, Emacs/W3, and EWW.)

• If Linux is so good, why don't I get Linux drivers on the CD/DVD with my sound/ethernet/video/etc. card?

  Actually, a few manufacturers do provide them, e.g. LinkSys for its ethernet cards.

  However, in general, we of the Linux community don't want hardware manufacturers to include Linux drivers. There are numerous compelling reasons:

  Such drivers almost invariably end up being (1) proprietary binary-only (because the manufacturers aren't yet used to an open-source world), (2) accordingly non-portable to all the others of the many CPU platforms Linux runs on, (3) liable to break unfixably as the kernel develops, (4) really poorly debugged, on account of no peer review, and (5) generally shoddy compared to what the community provides inside each and every kernel source archive. (Additionally, drivers get written for chipsets used in cards, not the cards themselves.)

  Case in point: Dell's shoddy, binary-only kernel drivers for their laptops. (They probably think we should be grateful for this sort of slop, and are no doubt doing this instead of cooperating with the developer community.)

  [Note: The above link at Dell Computer used to contain binary-only Linux kernel drivers for 3Com 3C575C ethernet cards and ESS Maestro sound chips. They were removed pursuant to investigation of their need to also provide source code under the author's specified licence, the GNU General Public Licence. Dell should be heartily commended for taking this step.]

  That is, superior drivers for common hardware are provided inside every Linux distribution, and it's rare for a component to not be supportable by them. In those rare cases, support is coming, via the fastest software development model in history.

• How do I find an ISP that supports Linux?

  Mu. (The question cannot be answered as posed.)

  The meaning of the term "support" in this context is hopelessly confused: The questioner may be intending to ask if any technological or legal obstacles exist for Linux users, or whether anyone has sample configurations for Linux, or whether the ISP will give any help at all with Linux access. However, the ISP's priority is to limit its support-staff costs, and the support manager is obliged to pessimistically assume "support" means "holding the user's hand, no matter how inept he/she is, and no matter how badly he's already mangled his/her system, until his/her problem goes away."

  He/she probably hires low-paid support staff who rely heavily on canned answers for pre-planned scenarios with a carefully limited set of user software. When you call and pose a question that has "support" and "Linux" in it, he/she envisions a bottomless demand for expert assistance with endlessly varying, hopelessly diverse and obscure software to demanding customers who've customised their environments to hell and back. His or her answer is therefore an automatic, emphatic "No."

  Which fundamentally means next to nothing.

  Consider as an example AT&T Business Internet Services, a high-quality global provider. You decide you want an account. Until recently, this required downloading Win32 or MacOS sign-up software, which took took down your credit-card details, made you "sign" a rather eyebrow-raising terms-of-service document (e.g., you agreed to be held liable for any e-mail purporting to originate from your address), had you pick a local dial-up number, and issued you an account name and password. Those account details could be subsequently used successfully on any operating system using the standard PPP protocol with PAP authentication — but few users realised that fact.

  These days, AT&T alternatively offers 100%-Web-based sign-up, usable from any SSL-capable browser, complete with Linux PPP instructions. (Other providers, such as the late Epoch Internet, have been known to agree to do customer sign-up entirely over the telephone, if you say you're not a Win32 or MacOS user.)

  But the point is that whether an ISP can be used with standard protocols has nothing at all to do with whether it claims it "supports Linux". Unfortunately, although the obvious avenue to determine the former is to inquire about technology (e.g., "Is PPP and PAP all that's required for your dial-up dynamic-IP service?"), you may have to ask ten people in a row before getting a straight answer, rather than brush-offs, non-answers, and wrong answers. There is no good solution, other than patience, courtesy, persistence, and skepticism — not to mention understanding the technology. (Ditto for xDSL.)

• What happened at Linux Gazette?

  In February 2010, five senior editors at the magazine including me complained privately to Editor-in-Chief Ben Okopnik about an editing practice he was indulging, that we felt was abusive towards our volunteer authors and made the magazine look bad in print. His response was to ignore our complaint, call it a "flamewar", kick out dissidents (including me), and imply in the magazine's next issue that I had somehow acted to injure the magazine's technical operations. That is wholly untrue; I continued to provide and fund all services (DNS, mailing list and MTA hosting, etc.) on a neutral basis. Ben ceased to use those gratis services entirely on his own initiative (and without notice).

  To refute some of the innuendo: 1. No, I didn't withhold any mailing list data, e.g., the only reason why Ben didn't collect lg-announce's subscriber roster is that he didn't ask (nor even mention desiring to move the mailing lists elsewhere — and that roster was private by his own policy). 2. The only reason the TAG mailing list had "a high level of spam" until mid-2009 is that Ben kept rejecting my advice of subscriber-only posting. When Ben changed the "subscriber policy" he speaks of, he set discarding of readers' mail, such that I had to fix that. 3. Far from fighting with Ben and then attempting sabotage, I presented my complaint (as did four other senior editors), was ignored, and sat back. I was thus surprised to be not only ejected but also thereafter singled out for personal, public maligning within the magazine's text.

  When the five editors' complaint was summarily dismissed as a "flamewar", I did temporarily set all listadmin passwords to a random string nobody knew, so nobody could take rash measures such as mass-deletion of entire mailing lists, or of their subscribers, or of their archives. Ben said he intended to "retaliate" for my placing that temptation out of everyone's hands: Thus, I guess, his public innuendo.

  (I would otherwise not have commented here, or anywhere else in public, on a dispute that should have remained internal to the magazine.)

Last modified: 2025-03-07

rick@linuxmafia.com

Copyright (C) 1995-2025 by Rick Moen. Verbatim copying, distribution, and display of this entire article (page) are permitted in any medium, provided this notice is preserved. Alternatively, you may create derivative works of any sort for any purpose, provided your versions contain no attribution to me, and that you assert your own authorship (and not mine) in every practical medium.

---

Top of page