Linux in the real world

Hubs don't tend to play well with DSL, and even switches are often less than fully functional, if there isn't a local router involved. The problem is largely with the way the network is set up on the ISP's side: they make things difficult for running multiple systems.

The problem that arises is that you only have one DSL account (I assume), and as such they only give you one IP address. A hub or (basic) switch thus can't help much because multiple systems plugged into a hub or switch will need multiple IP addresses.

You need either a NAT device of some kind or a router/gateway, so that the ISP only sees one device at your locality. This is true no matter what OS you're using.

To look for a package, do a search by what sort of software you're seeking. For instance, if you want a text editor, you could try searching for "editor". If you want a word processor, you could try "word processor". If you're looking for OpenOffice.org, you could search for it with something like "openoffice". The quotes shouldn't be necessary unless your have spaces in your search term.

The command to perform the search is [b]apt-cache search[/b]. For instance, if you're looking for a word processor, you might try this:

$ [b]apt-cache search "word processor"[/b]

If you're looking for OOo in particular, you might try:

$ [b]apt-cache search openoffice[/b]

If you find something in a list of search results that catches your eye, but you want to know more about it, you can use apt-cache for that as well. For instance, in searching for a word processor, you might come across abiword (my personal favorite word processor at the moment, when I have to look at Word documents and the like). To see more about abiword without installing it or searching the web, you could use the [b]apt-cache show[/b] command, like so:

$ [b]apt-cache show abiword[/b]

The output of that command is as follows on Debian Sarge/stable 3.1:

[i]Package: abiword
Priority: optional
Section: editors
Installed-Size: 6620
Maintainer: Masayuki Hatta (mhatta)
Architecture: i386
Version: 2.2.7-3
Replaces: abiword-common (<< 2.0.14-1), abiword-gtk, abiword-gnome
Depends: libatk1.0-0 (>= 1.7.2), libc6 (>= 2.3.2.ds1-4), libenchant1, libfontconfig1 (>= 2.3.0), libfreetype6 (>= 2.1.5-1), libfribidi0 (>= 0.10.4-5), libgcc1 (>= 1:3.4.1-3), libglade2-0 (>= 1:2.4.2-2), libglib2.0-0 (>= 2.6.0), libgtk2.0-0 (>= 2.6.0), libice6 | xlibs (>> 4.1.0), libpango1.0-0 (>= 1.8.1), libperl5.8 (>= 5.8.4), libpng12-0 (>= 1.2.8rel), libpopt0 (>= 1.7), libsm6 | xlibs (>> 4.1.0), libstdc++5 (>= 1:3.3.4-1), libx11-6 | xlibs (>> 4.1.0), libxft2 (>> 2.1.1), libxml2 (>= 2.6.16), libxrender1, zlib1g (>= 1:1.2.1), abiword-common (= 2.2.7-3)
Conflicts: abiword-gtk, abiword-gnome
Filename: pool/main/a/abiword/abiword_2.2.7-3_i386.deb
Size: 2340758
MD5sum: 27dd095ebe465f4b615483487c716f43
Description: WYSIWYG word processor based on GTK2
AbiWord is the first application of a complete, open source office
suite. The upstream source includes cross-platform support for Win32,
BeOS, and QNX as well as GTK+ on Unix.
.
This package contains the AbiWord binary built with GTK2.[/i]

It uses GTK, which annoys me a little, but at least the interface doesn't require Java like OOo's does.

The other apt command set, aside from apt-cache, that is likely to be of great use to you is apt-get. The apt-cache commands access a locally stored cache of data (thus the name) about available packages. The apt-get commands get data and packages from some other source, defined in your /etc/apt/sources.list file. These other sources might be a local apt archive mirror, a set of Debian install CDs, or an apt archive accessible through the Internet. I tend to use the hell out of the Internet-based archives, personally, though as I roll out more Debian systems at work (knock on wood) I'll want/need to set up an apt archive mirror on the LAN. In any case, as I said, apt-get is used to get data and packages from your defined apt archives (thus the name).

Finally, there's apt-source, which is used to get and install source packages, but we'll ignore that one for now: you can happily get along without that for a long time to come, most likely.

Once you've used apt-cache to figure out what package you want to install, you can use apt-get to install it. For instance, if you want to install abiword, you'd use the following command:

# [b]apt-get install abiword[/b]

The hash mark "#" at the beginning of that line indicates a command prompt that ends with "#" (for most Linux systems, that's the default end-character of a bash prompt when logged in as root), whereas the earlier use of the "$" character indicated that the dollar sign, instead, was the prompt ending character. This is to indicate that, for instance, apt-cache will run as pretty much any user, but apt-get requires administrative rights, and you must thus run it as root.

The apt-get command has a lot more capability than that, however. For instance, if you're [b]get[/b]ting from an Internet-based archive, you'll want to udate your local cache regularly (perhaps daily). To do that, you would (unsurprisingly) use the [b]apt-get update[/b] command, like so:

# [b]apt-get update[/b]

Yes, it's that easy.

The next thing you'll probably want to use is [b]apt-get autoclean[/b], to clear out any older package references in your local cache that are no longer relevant. Finally, there may be new versions available for packages you already have installed. If you're using Debian stable (currently stable is v3.1, aka Sarge), you will almost never find that there are new versions aside from security patches and similarly critical updates, and as such it's a very good idea indeed to check for new package versions with the stable release.

To check for, and install, package updates, you'd use [b]apt-get upgrade[/b]. No additional arguments are needed. As with install, update, and autoclean, above, you have to be root to use the upgrade command.

When you run [b]apt-get upgrade[/b], it will spit out a list of packages that can be upgraded to a newer version. It will then ask whether you want to continue with the upgrade. In general, you should probably just press the Enter key (the question will be followed by [Y/n], indicating that you can answer either with a "y" character or "n" character for "yes" or "no", with the capitalized Y indicating that it's the default so that simply pressing Enter will be the same as entering "y"). This allows it to install all the indicated package upgrades.

You can also choose "n", for "no", of course. If you see some package in there that will upgrade something you don't want upgraded (you'll surely know when you need to choose a "no": it's like the old comment about products without pricetags, where if you have to ask, it's too expensive). If you decide you want to upgrade some packages but not others, you can always learn how to "pin" certain packages, set them to not upgrade, and so on. This is a slightly more complex task than I intend to discuss right now, and is likely unnecessary for you. You can also simply refuse batch upgrades like that and, instead, upgrade one package at a time. Using [b]apt-get install[/b] on a given package can be used to upgrade it if there's an upgrade version available.

To remove packages (the opposite of installing them), you would use [b]apt-get remove[/b] (surprise, surprise), like so:

# [b]apt-get remove openoffice.org[/b]

If you want to get rid of all the configuration files used by that package, you'd use the --purge switch, like this:

# [b]apt-get --purge remove openoffice.org[/b]

I generally recommend doing that, unless you think you'll be reinstalling the package later and will want the same configuration options to apply (after you've changed preferences for the application in question).

That's pretty much it. To search for a package, use apt-cache search. To get more information about it, use apt-cache show. To update your cache, use apt-get update. To install a package, use apt-get install. To remove one, use apt-get remove (or apt-get --purge remove). Et cetera.

There's another tool of immeasurable value, called debfoster, that you might want to install.

# [b]apt-get install debfoster[/b]

As you install packages over time, occasionally run debfoster at the command line (just enter [b]debfoster[/b]). This is used to present you, one at a time, with a list of installed packages that you may want to either keep or remove. This is one way you can avoid having to memorize every single package you have installed on your system in case you might some day want to remove one of them.

It will give you information about each installed package that looks something like this:

[i]bastille is keeping the following 7 packages installed:
libbit-vector-perl libcarp-clan-perl libcurses-perl libdate-calc-perl libnetwork-ipv4addr-perl libunix-syslog-perl psad
Keep bastille? [Ynpsiuqx?], [H]elp:[/i]

To tell it to remove a package (as if you'd entered apt-get --purge remove), press the N key. To keep it, and cause debfoster to not ask you about it next time, either press the Y key or simply press Enter. To get more information about the package before you make a decision, use a question mark (hold down Shift and press the key with / and ? on it, just like you were typing a question mark in a word processor). If you want to skip this one, so that it's not removed but you are asked about it again the next time (something you'd do if you're not sure yet whether you want to keep it), press the S key. For help on how to use the debfoster options, you can use the H key.

All of these options are available to you, as you might notice from the above-quoted output from debfoster. That "[Ynpsiuqx?], [H]elp:" bit gives you a string of letters, plus an H for help, as options. The Help key tells you very briefly what each of those letters does. If you want to quit debfoster without continuing, such as when you've told it to remove something you don't want removed or to keep something you don't want kept, you can use the Q key, then you can start over by running debfoster again.

So, the most important stuff for a beginning Debian user:

apt-cache search
apt-cache show
apt-get update
apt-get upgrade
apt-get install
apt-get remove
debfoster

By the way, if you get results that are too damned long to read easily in your console window from [b]apt-cache search[/b], you might try piping it through a pager, such as less:

[b]apt-cache search editor | less[/b]

Between "editor" and "less" is the "pipe" character, also known as the "vertical bar" to Windows users. It's typically on the same key as the backslash. In bash, the pipe character is used to "pipe" one command's output through another command. Thus, if you want to send the output of a very long list of packages from an apt-cache search to a pager like less so you can scroll up and down through the list, you can use the pipe character to do that. I typed it with spaces between the pipe character and the words on either side, for clarity, but you don't necessarily have to have those spaces there.

I hope that helps.

Shortly before 1986 (1981-1985) I worked for DEC (Digital Equipment Corp.), and they had a few Unix versions -- Ultrix 32, and Ultrix 16. Later they had OSF/1 and True64, I believe. In any case, I can't imagine that there was an OS that was a cross between Unix and DOS at that time named "dec".

DEC had a number of other operating systems that don't really meet your description. What kind of hardware did this OS that you remember run on?

You could always try a minimal install of Debian from the very beginning, without using tasksel, aptitude, or anything else to install anything more than the base system during OS install. Then, when you can boot to a console, use apt-get to start installing the software you want to use. It's a little more advanced to do it that way than to let tasksel install whatever it thinks is appropriate for a "desktop" system (for instance), but it has always worked for me, and makes me much happier with the end product than I would be with some other schmuck's wildest guesses as to what a common end-user might like.

You should be able to run a Debian install, even with GNOME or KDE, on 8MB of RAM. If it's not running properly, your problem is something else.

When you've supported as many end-users trying to install Windows from scratch, or trying to resolve issues when hardware gets replaced, as I have, then perhaps you'll realize that Windows is at best no better than Linux for issues like this.

"[i]Aunt Betty and her quilt knitting friends will NEVER edit config files or compile source code.[/i]"
They'll never hack the Windows Registry, flash BIOS, or resolve driver conflicts through the Device Manager, either. The real difference is in the fact that it's harder to find vendors that sell complete systems with Linux pre-installed and pre-configured.

"[i]in some distros you have to be a geek even to find the documentation by using the CLI MAN commands[/i]"
Funny, I never realized you had to be a "geek" to type "man" and hit the Enter key. That's news to me. In any case, I'd rather have obscure documentation than none at all.

"[i]A while back I read a piece about some guy trying to print to a shared printer (on XP), it went on to describe his efforts digging through the documentation, setting up SAMBA, dealing with dependencies, etc.[/i]"
Funny, with Red Hat using KDE (for instance), I just click Control Center > System > Printing Manager and use the GUI printing configuration tool; this is very like the way Windows handles it. On Mandrake with KDE it's slightly more complex: I open the Mandrake menu and go to System > Configuration > Configure Your Computer > Hardware > Printers (which is made a little more obvious by the fact that the icon for Hardware shows a printer). It's similarly easy in GNOME. While I wouldn't throw a user who'd only used Windows previously into CLI printing configuration without waterwings, even it is not too difficult on distributions like Debian (my favorite).

"[i]The same thing goes for installing applications. In Windoze, you put the disc in, double click 'setup.exe' and it auto-magically does its thing.[/i]"
Jeez, that sounds pretty much like a pain in the arse. You have to have a CD? I just type "apt-get install [packagename]" in Debian, or "urpmi [packagename]" in Mandrake, or "yum install [packagename]" in Fedora Core, and it installs, dependencies and all. Even better, unlike with Windows applications, installing a new one using "apt-get install" doesn't tend to break other applications that use some of the same libraries (DLLs in Windows-speak).

"[i]Just because some functionality can be added in, doesn?t mean that is should be added in.[/i]"
Gee, I've been saying that about Microsoft for years, and they're still not listening to little old me. Next time you decide to say something like that, maybe you should have a look at the broken-ass mess of "features" that MS calls "Office" first.

"[i]Once your software is ?Aunt Betty proof?, and accomplishes its primary goal (i.e. the printer auto-magically prints), then you can make the advanced features available through either config files or plug-ins or whatever.[/i]"
Well, good. Then we're done. See above, re: KDE and GNOME configuration of printing. By the way, "config files . . . or whatever" is what's going on behind the scenes, just as Windows Registry settings, MIME type configuration, DLL dependencies, and so on are going on behind the scenes when you configure printers for Windows. This isn't "extra functionality" that gets added on later. This is what makes it all work in the first place. The point of using plain text config files rather than a Windows Registry is to make it easier for an expert to troubleshoot when Aunt Betty completely screws something up.

"[i]Make installers that include everything your software needs to run![/i]"
Ooh, bloat. Good plan. More complexity. Less secure software is fun. More broken software is fun. Oh, yeah, and even Windows doesn't do this as much as you think. Luckily, unlike Windows, major Linux distributions provide massive software package repositories that allow you to ignore things like dependencies and just say "install this thing", and it installs and automagically works. I'm guessing you haven't actually tried this stuff out, though.

"[i]If you can?t grab it off the shelf, rip off the shrink-wrap, and install it,[/i]"
Why would I want to do all that? I can just skip the "grab it off the shelf" and "rip off the shrink-wrap" steps and go straight to "install it".

The first automobiles were a hodgepodge of sometimes weird control systems. Aircraft were more complex and featured some tweaky ideas.

Computers speak binary, and if you think about it there are at least 6 levels of translation between the average user and the place where things are really happening. Everything inbetween is "virtural".

We have focused on a concept of performance because what has driven the development is largely the visual interactive part of the pardigm. And not the, "Here is a very usable system that has a word processor and email functions that is secure, robust, and self repairing." idea. If just 10% of the money and sweat that has been used to develop games was put into the second goal we would have an "Aunt Betty" system available for $9.99 at your local store.

Linux is not for the mainstream for the same reasons that learning to fly is not for the mainstream. Is Aunt Betty a pilot? Would you trust her in a cockpit, shoot, do you trust her behind the wheel of a car?

Linux is growing up. There will be wrong turns, incorrect decision, and problems. If you have the inclination join those who are trying it out. At least we do not have broken bones when our "follies" crash.

The world is becoming more complex, partially because of computers. What we dream today becomes our reality.