Radix Fidem Blog

You probably know everything on your computer is just zeroes and ones, grouped together in eights and sixteens, and so forth. You might know a bunch of folks type out lines of instructions (“code”) for computers which have to be converted into those ones and zeroes, in a process called compiling. You may have heard compiling stuff on Windows requires you to buy expensive software suites to do that sort of stuff. Maybe you know Open Source means the entire process from start to finish is wide open and free, and if you take a notion, you can compile your own software because it’s all part of the package. This is why I recommended you install the Developer Workstation package profile. You can take all those instructions people write and make it into zeroes and ones yourself.

On RHEL, you have the option of going out and finding that code from the sources and learning how to use the common build scripts. I do it all the time. But it’s a whole lot simpler chasing down all the extra piles of code you need if it comes pre-packaged. For RHEL, that means the RPM system. It’s not just a way to add packages someone else built for you, but you can build the packages, because the source is also neatly packaged. We call them source RPMs or SRPMs. Whenever you see a package which ends in the combination “.src.rpm” you have an SRPM.

Depending on what we are trying to build, we can get our SRPMs from several different places. For RHEL, the primary source is the official RHEL 6 workstation repository. Occasionally I also have to check the server repository. These two are updated regularly when something is fixed, so check to get the version with the highest number, which may mean the last in a long string of digits, dots, dashes and underscores. A much wider variety of packages come from the Fedora repositories. If you understand Fedora is the playground development system for the serious RHEL working systems, then it’s no surprise Fedora has more different kinds of stuff. You just have to be aware of which of the many versions of Fedora is closest to your version of RHEL. In this case, RHEL 6 is built largely from FC13, but parts of it appear to come from FC12. At any rate, if you don’t find what you need on the RHEL repositories, your second look should be “FC13 Updates” and “FC13 Everything” in that order. You can find a list of mirrors here simply select the one closest to you. For each one, it’s usually a matter of this layout on the servers:

..../fedora/linux/updates/13/SRPMS/
..../fedora/linux/releases/13/Everything/source/SRPMS/

In the case of CentOS and Scientific Linux, they will have their own mirrors and structure when 6.0 is finally released by each of them. They are built from RHEL 6 SRPMs, and the same FC13 SRPMs should work for things not included in the primary SRPM sets.

Today we are going to grab the package called “freetype,” which is a standard item from RHEL, so go to the Workstation SRPMs and download that package. We can’t install this from the browser, so prepare to do some commandline work. Just as a reminder, in Linux, all commands are case-sensitive. Most commands are in lower case, but there are switches (usually with a dash in front of them) which are both upper and lower case, so you need to pay attention to them.

Open your Terminal and su to root. Type cd so you are working from root’s Home. You can issue the command to install this SRPM from here, as long as you declare on the commandline the exact location of that RPM file (called the “full path”):

rpm -ivh /home/username/Downloads/freetype- [hit TAB]

In this case, we use the “i” switch for a straight install, and the “vh” allows you to watch it work. You can ignore the warnings about who owns the files. When it’s done, if you type ls you should see a folder named rpmbuild. This is where all SRPMs will be installed so we can work on them. There are several sub-folders underneath. We are primarily interested in the SPEC folder, so type:

cd rpmbuild/SPEC

If you ls here you’ll find a single file, freetype.spec. This is the building configuration file. The Freetype library is the run-time reference for the way fonts are rendered on your screen. We are going to change it by adding two features not enabled by default, called “bytecode hinting” and “subpixel rendering”. If you take a quick look at this SPEC file:

less freetype.spec

you’ll see some interesting instructions at the top. That “less” command is equivalent to the Windows file viewer command “more” — less is more. The second line of the file tells you how to add those features. You see we can add a couple of “switches” on the commandline of the build process to make it happen. But it’s not a simple swipe of the mouse because there is some extra verbiage and punctuation. So you can hit “q” to close this viewer and I’ll give it to you below.

The basic command is rpmbuild, with a switch  (-bb) telling it to build the software and create the RPM packages. We’ll stick in the additional switches from inside the SPEC file and finish with telling RPMbuild what it’s working on:

rpmbuild -bb --with bytecode_interpreter --with subpixel_rendering freetype.spec

Hit enter and it should take off and build it, displaying the details on the screen in your Terminal. When it’s finished, the prompt will come back and we have to go find those packages and install them. If you take a moment to look in your Terminal window, you’ll see up just a few lines where it says “Wrote /root/rpmbuild/RPM/i686/freetype….” Each of those packages comes from the same SRPM. That’s the common pattern, and it also tells you where to find the packages. The source is not just the library, but some other stuff, which includes the development libraries. We need the base package and the “devel” package. Now, let’s go there:

cd ../RPMS/i686

That double dot is simply shorthand for “the directory above this one.” We go up one layer, then down two into the place where RPMbuild drops the finished packages, in this case the i686 folder (x86_64 for 64-bit). We need to install them both at the same time. And because we already have a couple of RPMs by the same names installed right now, but crippled versions we want to replace, we’ll need to tell RPM to ignore that. Keep in mind, as you type this rather long command, you only have to type enough to distinguish between files and hit the TAB key for each file. What you see below were the ones I built for this tutorial:

rpm -Uvh --force freetype-2.3.11-6.el6.2.i686.rpm freetype-devel-2.3.11-6.el6.2.i686.rpm

That’s the “rpm” command, with the “U” upgrade switch, and as before, the “vh” switches so you can see what’s happening. Then we add the “force” switch because it won’t do it otherwise. Don’t abuse that option! You seldom have any need for it, but it’s there and it can really mess things up and make your system quit running if you don’t know what you are doing. The command is completed by listing all the packages, each separated by a space. Hit ENTER.

In order to see the effects of this altered font rendering package, we have to restart the display server. That means log out, then log back in, because logging out typically restarts the display server. It should make at least some noticeable difference in how fonts are rendered, mostly by tightening up and sharpening the characters so they aren’t so fat, fuzzy and dark.

Welcome to RPM Land. More extensive instructions can be found starting here.

There is a world of difference between Windows and Linux in file handling. Your immediate need is to understand the business of permissions. One of the fundamental security advantages in Linux is every file is owned by someone, and the owner gets to decide what happens to those files. Ownership, of course, is limited to those who have an identity on the computer in question.

First, let’s open that Terminal again. When first opened, by default your relative location in the system is your Home directory. So important is this, anytime you issue the command cd without saying specifically where you want to go, you’ll be sent here to the same place, Home. On the Linux commandline, we love our abbreviations. When you are typing things on the commandline, the symbolic reference to your Home is ~/ — that’s a tilde and a forward slash together in that order.

Let’s see what’s there. Type ls (lower-case LS). You get a column list of item names, in different colors. Blue is what marks a directory, which may or may not have anything inside it. You should have at least Desktop and Downloads, and mostly likely some other standard folders such as Documents, Music, Pictures and Videos. Not so different from Windows in that sense. You may not see any files, which show up as black. That is, unless you’ve been playing with the color scheme in the Terminal menu system. But there isn’t much information about the folders and files. Who owns them?

Type “ls” again, but with an added switch, which asks for the full length version of information about each item: ls -l. You should see something like:

drwxr-xr-x. 2 ed ed 4096 Jan 5 08:42 Desktop
drwxr-xr-x. 2 ed ed 4096 Jan 7 14:33 Documents
drwxr-xr-x. 3 ed ed 4096 Jan 9 17:40 Downloads

That first column is all about permissions. I’m not going to duplicate the good work of others who write better than I. Allow me to refer you to this tutorial on Linux Permissions. You’ll need to read it if you don’t understand what we’ve done so far. Go ahead, we’ll be here when you get back. You might want to bookmark the home page of that website if you find yourself lost already in Linux. The author covers a lot of basics in very simple explanations.

For now, I want you to learn how to do this in the GUI. If you click the “Home” icon on your desktop, you get a window reminiscent of the ancient Windows 95 default. If you like that, fine, but it won’t help you learn much about maintaining your own file collection. So in the open window, go to the Edit menu, and select Preferences, the last item. You should see a configuration dialog with many tabs at the top. In the first tab, hit the drop-down box where it says “Icon view” and change it to “List view.” I recommend you checkmark “Show hidden and backup files.” On the Behavior tab, checkmark “Always open in browser windows.” My personal preference is also to select the radio button for single-click, because I like my file browser to act like a web browser; everything is a link which requires only one click. Finally, go to the “List columns” tab and add a checkmark to the box at “Permissions.”

Hit the “Close” button; the choices are automatically saved. Now close the file manager window. Open it again, and the changes should have been applied. Of the far right side of each line you see those permissions again. Right click on any item, folder or file, and click “Properties” at the bottom. Click on the “Permissions” tab and you get a detailed management page for the item in question. You’ll see the labels explaining all that stuff you saw in your Terminal window. Change those permissions and it changes what shows up in the file browser window in that permissions column on the far right. Note: All folders must be marked with the “execute” bit for you, or you can’t see what’s inside.

You’ll discover if you transfer any files from a FAT file system (like thumb drives or memory sticks), NTFS (the current Windows default), or most backup CDs, every file has all the execute bits set. That’s how Linux interprets Windows default file settings. It’s not a big problem in every case, but it does conflict with the Linux defaults. When copying files from those other media into your Home folder or anywhere else on your Linux machine, be aware of this. If the file is not supposed to be an executable script or binary, change those permissions to take away the execute bit (but not on folders). It’s laborious if you have a lot of files, but it’s just one of those things.

Now, about ownership: You’ll notice the files belong to you as your username, listed twice. That’s because there is the user as a person, and you belong to a group by the same name. You may have picked up on that from the tutorial I linked above. The root account can change ownership of any file, but you get to set group membership for any group your user account belongs to.

Linux and Unix folks will gravely warn you about messing around too much while logged in as root. On your system, root owns everything, and can delete the entire system with just a few keystrokes. So we protect our root access with hard passwords and don’t use the root login except when necessary. Never connect to the outside Net as root unless absolutely necessary, as this exposes your machine to someone else using some unknown security hole to masquerade as root on your machine.

In the last lesson, we copied a bunch of codec files into the library file system. Those files are still owned by your user account. It’s a low-level security threat to leave them that way, among other things. You could accidentally delete them from your user account. Suppose you were online and someone managed to use one of those as-yet undiscovered security holes, and grabs control of your system using your user account. They will have only your user account’s permissions to do things. Those files are not protected. As a standard practice, anything which is used by the system should belong to root. Let’s fix that situation.

Open the Terminal and login as root. Then execute this command:

chown -R root:root /usr/lib/codecs/

That’s change ownership, using the -R switch for recursively (diving down into the folder and everything inside it), to make root the owner of all those codec files we moved there.

One final note. We got those files from a standard source in the Linux world. Unless someone packages them for your particular Linux version, that’s the way to get them. Right now, no one makes an RPM for RHEL 6 containing those codecs. If they did, we might be able to download and use that RPM file, and all would be well. But you should avoid adding multiple third-party repositories. ATrpms does things one way, and the others do things differently. If you add other repos which cover the same kind of packages, there will surely be conflicts, and Yum won’t be able to resolve them. Most of those third-party repos don’t work together, at least not very well, because the people who run them won’t, for whatever reason. If you use ATrpms, you might be able to take individual packages from other places, but even that could be dicey. It’s the nature of the beast.

Orient yourself to the desktop. The main menu system is in the upper left-hand corner. In the upper right is the notification area (“Systray”). On the lower toolbar, the left is where the open windows are listed, and the lower right is an iconic representation of multiple desktops with your desktop “trashcan.”

Windows does have a way to create multiple virtual desktops, but it’s not simple and many say it’s pretty cranky and unstable. I see this as a serious hindrance, because I’m often working with six or eight application windows open, and that’s too many for one desktop. In Linux, multiple desktops are the default. Right click on that display and you’ll see an option to configure preferences. Select that and you’ll see you can have even more desktops, and can configure them to display stacked in rows. Unless you make that bottom toolbar wider, it usually works best to keep them in a single horizontal row. But get used to the idea of having multiple desktops, because it allows you to organize the way you use Linux to get work done, by grouping open windows according to your work pattern on different desktops. To switch between the windows simply click on the miniature display at the lower right for the desktop you want, noticing as you do the open windows on each are thumb-nailed.

You make the toolbars wider by right-clicking on any blank space and select “Properties” to see the options. In the second tab, you can change the colors or even use a background graphic. You can also elect to add other nifty applets to the toolbars, if you look around in the context menu system. The GNOME desktop is loaded with similar features. Play around and get to know things. Setting more of the features at once comes by clicking the “System” menu, then “Preferences” followed by “Appearance”. You can change the window frame style, the color scheme, and the fonts. We’ll do more with fonts — a whole lot with fonts — later. You can change the wallpaper on your desktop by right-clicking on the wallpaper itself.

The context menu is the same idea as in Windows. For example, I prefer the clock in 24-hour mode, so I right-click on the time display and set the properties. You can get a reasonable weather report right next to your date and time display, once you tell the applet where you want the report sited. Explore; you’ll find it. You can also find the screensaver and power usage settings under the System > Preferences menu. Further, check the item marked “Sounds” — if you like audible cues coming from the interface, you’ll need to enable them.

One of the first things you need to watch for is a bright orange icon up in that notification area, looking like an explosion. That’s the update notification. It will require you to use the root password because it means changing the system. Click on it, read and follow the instructions. Let it do what it wants to do. If it lists an update using the word “kernel” you’ll have to reboot. This was actually one of the first updates after Red Hat released this version, so if you don’t see that notification within the first fifteen minutes, we’ll need to take action later to wake it up. It’s rare when you install any variant of Red Hat something or other without updates already waiting.

When you first install RHEL, it’s pretty light on desktop applications. They can be added easily enough, once you know what sort of things you find missing, and what RHEL calls them. Click on the “System” menu, then “Administration” and find at the top “Add/remove software.” You’ll have to use your root password again, but it should open a window and display what’s available. You may find the logic behind the layout a little hard at first, but you’ll get used to it.

In Linux generally, whoever supplies the operating system typically offers a wide array of software for it. For example, rare is the distribution of Linux which doesn’t have OpenOffice (or one of the variants, LibreOffice and Go-OO). RHEL 6 and clones call it OpenOffice, and it’s the latest, version 3.2.1 (at this writing). It’s not installed under the Software Development Workstation profile, but you can probably find it in the software installer utility and add it. The system automatically takes care of matching up the dependencies. Because it’s such a large package, it will take awhile.

Also, look for something called “alacarte” — that’s the package you need so you can edit the menus. We’ll need that later. Tomorrow we’ll secure Firefox, the one thing most people want to use first.

Please get a copy of the Installation Guide from this page; I recommend you keep a copy of the PDF version. Scan through it, because the visuals will help you understand what to expect.

Now, drop the DVD into the player and reboot.

The first thing you may encounter is display troubles. Some Intel chipsets cause the installer to freeze. If the entire screen goes black and stays that way, then try again. Reboot from the DVD and at the first screen which displays, select “Install with basic video driver.” This should get you past that well known problem. You’ll get a very primitive text-mode display for the next few steps, but at some point, if RHEL can use the graphics chips at all, it should give you a graphical display of some sort later.

The next issue is a very long pause while RHEL checks the hard drives available, as well as the networking environment. Depending on the circumstances, the installer will pause for an awfully long time, with the message about initializing the hardware. Be patient. As long as the screen is not simply blank, it’s okay.

The next step is the DVD media test. Run it the first time you use that particular DVD. RHEL is running a check to make sure it can read the entire disk. If you use the same disk again on another machine, it shouldn’t be necessary to scan it again, and you hit the tab key, highlight the word “skip” and hit enter. Either way, when it asks if you want to check another, select “continue” because we don’t have another.

This is where the installer’s hardware detector, called Anaconda, tries to pull up a graphical display if it hasn’t already. As with Windows, the “display” includes keyboard and mouse. Make sure you select the proper keyboard layout and default language. Then, it will check the hard drives again. RHEL will ask you if you have any specialized drives, which is highly unlikely. Just select “Basic storage device.” Sometimes RHEL will think it has found something odd and will throw up a warning about reinitializing the drive. There’s nothing you can do, really, so let it go ahead and do so. Then select “fresh install.”

Next is some networking configuration. Unless you have a LAN with some internal domain name, simply give your computer a name meaningful to you, a single word using all lower case letters. Unless you connect to the Internet via dialup, click that button for “Configure Network.” It will show you what RHEL has found possible so far. The main point here is to put a checkmark in the box marked “Connect Automatically.” This way you won’t struggle to configure it later, wondering why it’s not connecting. If you are limited to dialup, you cannot configure that here.

Select your timezone; the list is alphabetical by well known cities in your timezone. Here in the US, that’s choices like New York (Eastern Time), Chicago (Central Time), Denver (Mountain Time) or Los Angeles (Pacific Time). For the rest, just hunt around until you see something close to you. Unless you know your computer’s internal clock is set to UTC (Greenwich Mean Time), uncheck the box.

Next is where you enter your root password we discussed in the previous lesson. At some point later you will create your user account. Two or more letters for the username is fine, but really long names can be hard to type, so keep it simple.

Then we come to drive selection. If you have more than one hard drive in your computer, this can be pretty complicated. All the more so if they aren’t connect via the same bus. Some computers, like mine, can run both the older PATA drives with the flat wide gray cable, and SATA drives with the slender cable. Most operating systems will default to booting from the PATA drives, which is fine for the most part. You get to select which drives will be installed and which should be simply mounted to preserve the data. If anything is going to go wrong, this is most likely where it will be. It works best if all you have is a single drive large enough to run Linux, or the drives are of the same type. Again, I highly recommend you devote all the drive space to RHEL. Dual booting is dicey for Linux newbies, and RHEL doesn’t make it simple. But RHEL can figure out how to configure and use the drives just fine by itself if you don’t complicate things. Once you make up your mind, it will format immediately.

When RHEL asks you to select an installation profile, I am going to recommend you choose the “Development Workstation” — that’s Desktop plus the means to compile software. If you don’t have at least 80GB, your system may be too small for anything but the basic Desktop profile. Otherwise, we will need to build some packages soon, a lot of them if you have special needs. We’ll cover that later. RHEL 6 comes rather stripped down, even when you install the full set of development libraries, and you will inevitably build some libraries before you build the packages which rely on them. Once you finish choosing your profile, RHEL will write the packages to the hard drive. Take a break, because this won’t be quick at some 1400+ packages.

If all goes well, the installer will eject the DVD tray and wait for you to acknowledge the need to reboot. Shortly you will see a nice boot screen animation, then the welcome screen with some notes you should read. Next is the Linux equivalent of the “click-wrap” license. You can read it, but it simply declares the software could not possibly have a warranty, and if you play with any of the source code and make changes, you have to share it back with the community.

For users of the official RHEL release, you can enter your login credentials here, and connect to the Red Hat Network (RHN) for updates and so forth. Both CentOS and Scientific Linux (SL) skip that. Eventually you will have a chance to login to your user account.

Welcome to Linux.