Working in the X Environment This chapter shows you how to begin working productively in the X environment. It describes how to: Open a second xterm window. Move windows; raise windows to the front of the display; convert windows to icons. Close an xterm window. Start other clients in convenient places on the display. Run clients on remote machines. Customize a single client process using command-line options. Specify alternate default characteristics for a client using resource variables. At the end of the last chapter, you should've had the X server, the first xterm window, and the mwm window manager running. The current chapter illustrates some of the system's basic capabilities so you can begin working more productively. This chapter shows you how to: Open a second xterm window. Use the pointer to affect windows on the display. Iconify, deiconify, maximize, raise, move, resize, and close windows using the pointer on the mwm frame. Close an xterm window from the command line. Start additional client programs, on both local and remote machines. Organize the display. This chapter also introduces some basic ways to customize X clients to better suit your needs. windowscreating Once you focus input on the first xterm window, as described in Chapter 2, you can enter commands to open other client windows. For example, you can open a second xterm window by typing this command at the prompt in the first xterm window: xterm (terminal emulator)multiple xterms % xterm After a few moments, a second xterm window will be displayed on the screen. As we'll see later in this chapter, you can specify the location for a new window using a command-line option (or in many cases using a resource variable stored in a file in your home directory). If you don't specify position on the command line (or in a resource file), by default mwm automatically places new windows offset from the upper-left corner of the screen, as shown in Figure 3-1. If you start multiple processes in a row, the windows will be placed progressively further from the upper-left corner (towards the opposite corner), so that no window completely overlaps another. windowsplace using pointer interactivePlacement You can customize mwm to allow you to place new windows interactively using the pointer. This modification is performed by setting a resource variable called interactivePlacement to a value of true. See Chapter 13 for instructions on modifying mwm. See the mwm reference page in Part Three of this guide for more information about interactivePlacement.

The new xterm window displays a prompt from whatever shell you are using. In this case, the new window is running the UNIX C shell. Notice that the second window's frame is a dark gray, indicating that input is focused on it. The first window's frame has changed from dark to light gray; it no longer has the input focus. It's important to be aware that when you start a new window (and click-to-type focus is being used), the new window automatically takes over the input focus. (Note that the colors may vary according to system defaults.) windowsswitching between In the default mwm configuration, to switch back and forth between windows you must move the pointer from one window to the other and click the first button. Notice that if you are working with a stack of windows that overlap, selecting a window as the active window automatically raises that window to the top of the stack (in effect, the front of the display). Whatever you type will appear in the window with the highlighted frame. Try starting a command in both windows. For example, start up vi or another text editor in the second xterm window. Notice how you can switch back to the first window to type a new command, by moving the pointer and clicking--even if you leave vi in insert mode or some other command in the process of sending output to the screen. Whatever process was running in the window you left will continue to run. If it needs input from you to continue, it will wait. windowsmultiple You must always switch focus to work with multiple windows. However, mwm has complicated matters by placing the second xterm window in Figure 3-1 in a very inconvenient place. The second window overlies the first window and almost completely obscures it. Windows commonly overlap on the display. The window manager allows you to change the position and size of windows so that you can work effectively in such situations. The primary window management tool mwm provides is the window frame. The section “Managing Windows Using the mwm Frame,” later in this chapter, shows you how to perform these functions simply by using the pointer on various parts of the frame. But before we can learn to perform these window management functions, we need to learn more about pointer actions. pointerusing As explained in Chapter 1, the cursor on the screen follows the pointer's movement on the desktop. Move your pointer around on the desk to get used to this. Keep in mind that different pointers respond differently. If you move to another display, the screen cursor may move more quickly or slowly than the one you're used to. The xset client (described in &cmtf14;) lets you modify pointer behavior. buttonpointer You use the pointer to indicate a graphical element on the screen, such as a window, icon, or command button. Most pointers have three buttons. For our purposes, we'll refer to these buttons as first, second, and third, where the first button is the leftmost on the pointer, the second is in the middle, and third is the rightmost. Keep in mind that “first” is a logical distinction made by X, not a physical one. The first logical pointer button generally corresponds to the leftmost button on the pointer. (Thus, in some contexts you may find the buttons referred to as left, middle, and right.) However, X allows you to change the correspondence of logical and physical buttons. For example, you can reassign the first logical button to be the rightmost button on the pointer. A lefthanded person might opt to reverse the order of the buttons. You remap pointer buttons using a client called xmodmap, which is described in detail in &cmtf14;. By placing the pointer on a particular element and then performing some button action (and possibly a pointer motion), you can invoke a variety of commands. The types of button actions and pointer motions you can perform are: Click clicking buttons buttonclicking To click a button, you press the pointer button down and release it. A click is a rapid action; there is no pause between the press and release. A double click is two full clicks in succession, with no pause between clicks. A triple click is three clicks in succession. Press buttonpressing To press a button, you push the button down and hold it down. Release buttonreleasing After pressing a button down, you release it by letting up on the button. Drag pointerdragging item with dragging items To drag a graphical object (such as a window or icon) from one location on the screen to another: place the pointer on the object; press one or more pointer buttons; move the pointer to another location (dragging the object); and release the button(s). Keep in mind that some commands or actions are invoked by a simple click on a particular graphic element, as illustrated by mwm's click-to-type focus. Alternatively, some actions require a button press and pointer motion (i.e., dragging). When dragging is used to move an object, the actual object does not appear in the new location until you complete the movement and release the pointer button. Instead, you appear to drag an outline representing the object. When you release the pointer button, the actual object appears in the new location. This effect is illustrated in the section “Moving a Window,” later in this chapter. Dragging is also commonly used to change the size of a window. Again, an outline indicates that the window's size is changing. When the outline approximates the size you want, you release the pointer button and the actual window is redrawn using the selected dimensions. The following sections describe how to perform the most basic window management functions, which require you to use the pointer in the ways we've discussed. pointerusing mwm (Motif window manager) Figure 3-2 shows an xterm window “framed” by mwm. The window frame itself and several features of it are tools that allow you to manage the window using the pointer.

command buttonsMinimize The wider top edge of the frame is the titlebar. The titlebar is composed of several parts including a title area (displaying the name of the application) and three command buttons (Minimize, Maximize, and Window Menu). Notice that whenever you move the pointer into the titlebar, the pointer changes to the arrow cursor. Though it's not apparent from Figure 3-2, you can perform most window management functions by using the pointer on various parts of the frame. titlebar The following sections explain how to iconify, maximize, move, raise, resize, and close windows using the frame. Chapter 4 describes menu items and keyboard shortcuts that can be used as alternatives to the frame. These are important when a window's frame is obscured by other features of the display. Chapter 4 also describes additional functions provided by mwm menus. windowsiconifying Minimize command button command buttonsMinimize iconifying windows As discussed in Chapter 1, an icon is a symbol that represents a window in an inactive state. There are many circumstances in which it might be desirable to iconify a window: To prevent yourself from inadvertently terminating a window, as in the case of the login xterm. While running a program whose progress you don't need to monitor; if a window is tied up running a process and you don't have to see it, the window is just taking up space. If you are only using an application occasionally. For example, you might be running the xcalc calculator program, but only using it every so often. If you want to use several application windows, but only display a few at a time; this arrangement can be somewhat less confusing than a display crowded with windows. Having some windows iconified also frees you from constantly shuffling the stacking order. The Minimize command button on the mwm frame allows you to convert a window to an icon (iconify it). The Minimize button appears immediately to the right of the title area and is identified by a tiny square in its center. To iconify a window, use the following steps: Place the pointer within the Minimize command button. The pointer simply has to rest within the button's outer border, not within the tiny square identifying it. Click the first pointer button. The window is iconified. Figure 3-3 shows a window being converted to an icon in this way. By default, icons are displayed in the bottom left corner of the root window. mwm can also be set up to place icons in another location, to allow you to place them interactively using the pointer, or to organize icons within a window known as an icon box. In &cmtf13;, we'll discuss the specifications necessary to set up an icon box.

Minimize command button command buttonsMinimize windowsdeiconifying deiconifying windows iconsconverting to windows To convert an icon back to a window (deiconify it), place the pointer on the icon and double click, using the first pointer button. The window is redisplayed in the position it appeared before it was iconfied (and is also raised to the top of the stack). Between the first and second clicks, you'll probably notice that another small window is displayed for an instant above the icon. This window is actually a menu, called the Window Menu. (This menu can be displayed from a window or from an icon and can be used to invoke several window management functions on the window or icon. We'll discuss the Window Menu in more detail in &cmtf04;.) Be aware, however, that if you pause too long between the two clicks in deiconifying a window, the second click will not be interpreted and the icon will not be converted back to a window. Instead the Window Menu will remain on the screen, as in Figure 3-4. Notice that in addition to displaying the menu, clicking has caused the icon image to change in appearance. The icon label is wider and the label and the frame surrounding the icon are highlighted. These changes indicate that the icon has the input focus; thus the icon will interpret subsequent keystrokes as window manager commands. Notice also that the first item on the menu, Restore, is surrounded by a box, indicating that it is available for selection. Restore means to restore an icon to a window (or, as we'll see, a maximum size window to its original size). When the Window Menu0 is displayed above an icon, you can convert the icon to a window by placing the pointer on the Restore menu item and clicking the first pointer button. (Whenever a Window Menu item is boxed, you can also select it by pressing either the Return key or the space bar.)

If you want to remove the menu without invoking a command, simply move the pointer off the icon and menu and click the first pointer button. The menu will be removed; however, the icon will retain the input focus--the label and border will remain highlighted--until you focus input on another window or icon. buttoncommand command buttonsMinimize Minimize command button command buttonsMaximize windowsminimizing windowsmaximizing Maximize command button windowsenlargening Maximizing a window generally means enlarging it to the size of the root window. (In some cases, a client application may specify its own maximum window size and maximizing will produce a window of this size.) This action can be performed using the Maximize command button, which is located to the right of the Minimize command button (in the upper-right corner of the window). The Maximize command button is identified by a larger square in its center. It allows you both to enlarge the window to the size of the root window (or to the maximum size the client allows), and once it has been enlarged, to convert it back to its original size. To maximize a window, use the following steps: Place the pointer within the Maximize command button. The pointer simply has to rest within the button's outer border, not within the square identifying it. Click the first pointer button. The window is maximized, as illustrated in Figure 3-5.

The large window should function in the same way it did before it was maximized. Theoretically, you can maximize an xterm window to have a single, very large terminal screen. However, be aware that certain programs you may run within an xterm, such as the vi text editor, do not always work properly within a window of this size (even if you've used the resize client, as described in &cmtf05;). The Maximize function is more safely used with an application that displays a graphic image or performs a simple function, such as xclock. Also, some client programs that do not support resizing, such as the Release 3 version of xcalc, cannot xcalc (calculator)and resizing windows be maximized correctly. In the case of xcalc, the frame surrounding the calculator application is maximized, but the actual calculator remains the same size. The Maximize button is a toggle. To convert a maximized window back to its original size, click on the Maximize button again with the first pointer button. The Restore item on the Window Menu will also perform this function. Maximize command button windowsraising iconsraising We've already seen the necessity for raising windows on the display: frequently windows overlap. In order to work with a window that is all or partially covered by another window, you'll want to raise it to the top of the window stack. Using the default mwm, you raise a window with the following steps: Place the pointer on any part of the window frame, except the three command buttons (Minimize, Maximize, and Window Menu). Click the first pointer button. The window is raised to the top of the stack. When you are using explicit (click-to-type) focus, this click also selects the window to receive input, i.e., makes the window the active window. Once you have raised a window to the top of the stack, you should be able to enter input and read output easily. Windows may obscure icons on the display. (mwm does not allow one icon to obscure another.) If an icon is partially visible under a window, you can raise it using the following steps: Place the pointer on the obscured icon. Click the first pointer button. The icon is raised to the top of the stack. Figure 3-6 illustrates an icon being raised in front of a window. Notice that in addition to being raised to the top of the window stack, a menu (called the Window Menu) is displayed over the icon. (This menu can be displayed from a window or from an icon and can be used to invoke several management functions on the window or icon. We'll discuss the Window Menu in more detail in Chapter 4.) To remove the menu, move the pointer off of the icon and menu and click the first button. Notice also that the icon image changes in appearance when you raise the icon to the top of the stack (as it did when we paused between the double click to deiconify). The wider label and the highlighted label and frame indicate that the icon has the input focus. (Remember, when an icon has the input focus, it will interpret subsequent keystrokes as window manager commands.) Even when you remove the Window Menu, the icon will retain the focus (and remain highlighted). When you direct focus to another window (or icon), the icon label will become normal again.

windowsmoving title areamwm frame mwm (Motif window manager)frame;title area movingwindows titlebarand moving windows In many cases you'll want to move a window from one location on the display to another. The largest part of the titlebar is known as the title area, primarily because it displays the name of the application. The title area allows you to move the window, using the following steps: Place the pointer within the title area. The pointer changes to the arrow cursor. Press and hold down the first pointer button. Move the window by dragging the pointer. Figure 3-7 shows a window being moved in this way. When you begin to move the window, the pointer changes to a cross arrow pointer and a window outline appears. This outline tracks the pointer's movement. In the center of the screen, a small, rectangular box also appears, displaying the x and y coordinates of the window as you move it. Drag the cross arrow pointer with the window outline to the desired location on your screen. Release the first pointer button. The window will move to the selected location.

With the default configuration of mwm, moving a window also selects that window as the active or focus window. iconsmoving movingicons Moving an icon is similar to moving a window. To move an icon: Place the pointer on the icon. Press the first pointer button. Move the icon by dragging the pointer. Figure 3-8 shows an icon being moved in this way. When you begin to move the icon, the pointer changes to a cross arrow pointer and an icon outline appears. This outline tracks the pointer's movement. Drag the cross arrow pointer with the icon outline to the desired location on your screen. Release the first pointer button. The icon will move to the selected location. With the default configuration of mwm, moving an icon also selects that icon to receive the input focus.

mwm (Motif window manager) windowsresizing resizing windowsusing pointer One of the most distinctive and useful features of the mwm window frame is not at all obvious. The entire frame (other than the titlebar--i.e., the title area and command buttons) is designed to allow you to resize the window using the pointer. Notice that the frame is divided by small lines into eight sections: four long borders (two horizontal and two vertical) and four corners. Figure 3-9 shows these sections of the window frame. You can resize a window horizontally, vertically, or simultaneously in both directions. Resizing is a bit trickier than any of the other window management functions we've tried, since the way you move the pointer determines the size of the window. It will probably take some practice. If you place the pointer within a window and then move it into one of the long horizontal or vertical borders, you'll notice the pointer changes to a new shape: an arrow (pointing toward the window border), with a short line perpendicular to it. This short line represents the window border. Try moving the pointer in this fashion in one of the windows on your display to get a better idea of what the pointer looks like. If you move the pointer from within a window into the outer border at one of the corners, the pointer will become an arrow pointing diagonally at a small corner symbol, as pictured in Figure 3-10. Figure 3-11 shows all of the possible resize pointers.

Once the pointer changes to one of these shapes, you can move the border (or corner) of the window. Resizing from one of the long borders only allows you to change one dimension of the window: a horizontal border can only be moved up or down, changing the height; a vertical border can only be moved left or right, changing the width. Resizing from a corner offers the most flexibility. You can move a corner in any direction you choose, changing both dimensions of the window if you want. For example, you can drag the lower-right corner of a window down and to the right to enlarge the window in both dimensions. You determine the size and shape of the window by choosing the border or corner you want to extend (or contract) and moving it the desired amount using the following steps: Move the pointer from within the window to the border or corner you want to move. The pointer changes to one of the symbols pictured in Figure 3-11. Press and hold down the first pointer button and drag the window border or corner in the direction you want. As you resize the window, an image of the moving border(s) tracks the pointer movement. Also, in the center of the display, a small rectangular window shows the dimensions of the window as they change (in characters and lines for xterm windows, in pixels for most other clients). Resize the window as desired. Release the first pointer button. The window is redisplayed in the new shape. (The border image and window geometry tracking box disappear.) Figure 3-12 shows a window being “stretched” from the lower-right corner. Note that resizing an xterm window will not change the dimensions of the text currently in the window. If you make the window smaller, for instance, some of the text may be obscured. On most operating systems, this should not be a problem. As you continue to work, perhaps starting a text editor, the text will be adjusted to display in the newly sized window. Problems are more likely to occur if you resize a window during a text editing session. It's likely that the text editing program will not know about the window's new size and will operate incorrectly. To solve this problem, simply quit out of the editor and start another session.

If your resized xterm window does not seem to know its new size, you may be working with an operating system that does not support terminal resizing capabilities. Refer to the discussion of the resize client in &cmtf05;, and to the resize reference page in Part Three of this guide for alternative solutions. menusWindow Menu (mwm) Window Menucommand button The command button on the left side of the titlebar is used to bring up the Window Menu, which provides seven items that can be used to manage the window and its icon. &cmtf04;, describes how to bring up the Window Menu and invoke its various functions. This command button also has another function. Double-clicking the first pointer button on the windowskilling;with Window Menu button killingwindows Window Menu command button kills the client program and closes the window. Be aware, however, that like other methods of “killing” a program (such as the xkill client), double-clicking on the Window Menu button can adversely affect underlying processes. Refer to the section on xkill in &cmtf08;, for a more complete discussion of the hazards of killing a client and a summary of alternatives. You can customize mwm so that double clicking performs no function by setting a resource variable, wMenuButtonClick2, to false. See the sections “Setting mwm Resources” and “mwm-Specific Appearance and Behavior Resources” in Chapter 13, and the mwm reference page in Part Three of this guide for details. windowscreating menusWindow Menu (mwm) xterm (terminal emulator)terminating terminating xterm window killingwindows windowsexiting xterm windowsclosing exitingxterm window When you are finished using an xterm window, you can remove it by typing whatever command you usually use to log off your system. Typically, this might be exit or Control-D. We'll close the second xterm window in Figure 3-13 by typing exit. Notice that when we terminate the second xterm window, the first xterm window takes over the input focus. When explicit focus is being used and a window is terminated, the input focus reverts to the window that previously had the focus. Be aware that terminating the login xterm window (the first xterm to appear) kills the X server and all associated clients. (If xdm is running X, the server will be reset but only after all client processes have been killed.) Be sure to terminate all other xterm windows before terminating the xterm login window. Also, if you are in an editor such as vi, be sure to save your data before you terminate the window. windowsiconifying iconifying windows In fact, it may be wise to iconify the login window and use other xterm windows instead, so that you don't inadvertently terminate it. Remember: you iconify a window by placing the pointer on the Minimize command button on the frame and clicking the first pointer button. If you are worried about typing Control-D (end-of-file) accidentally and you normally use the C shell (csh), you can enter: % set ignoreeof in the login window. Then typing exit becomes the only way you can terminate the window. Note that some C shell implementations have an autologout variable, which will automatically terminate the shell if there is no activity for a given period of time. If your C shell supports this feature, be sure to disable it in the login xterm window using this command: % unset autologout As an alternative to entering the command used to log off the system, you can also terminate an xterm window by selecting Send HUP Signal, Send TERM Signal, Send KILL Signal, or Quit from the xterm Main Options menu. (These menu options send different signals to the xterm process. Depending on what signals your operating system recognizes, some of the options may not work as intended. See Chapter 5 for more information.)

The mwm window manager also provides several ways to remove an xterm or any other client window, among them double-clicking on the Window Menu command button, as described previously. Additional methods are described in &cmtf04;. clientsstarting additional Now that you know the basics of managing windows, you can start other X clients just like you can start another instance of xterm. The following sections describe how to open more client windows, place them on the display in convenient positions, and take advantage of X's networking capabilities by running clients on other machines. To start a client, at the command-line prompt in any xterm window, type the name of the client followed by an ampersand () to make the client run in the background. For example, by typing: % oclock oclock (analog clock) you can start a clock application. The clock will appear in the upper-left quarter of the screen. With non-rectangular windows like oclock, a titlebar appears to float above the window. You can then drag the oclock window to a more convenient location--say the upper-right corner, as in Figure 3-14. (Remember, to move a window, place the pointer on the title area, press the first button, drag the window outline to the desired location, and release the button. Notice that the outline is rectangular, even if the oclock window isn't!)

Though moving windows on the display is fairly simple, manually positioning every window is not particularly convenient. For most clients, X provides a way to specify a window's location (and also its size) automatically--using an option when you run the command. A window's size and position is referred to as its geometry and you set these attributes using the -geometry option. The use of this option is discussed in the section “Window Geometry: Specifying Size and Location,” later in this chapter. Unfortunately the developers of oclock neglected to provide an easy way to remove it. For instructions on removing an oclock window, see &cmtf08;. oclock (analog clock)removing killingclients windowsremoving clientsoptions command-line options optionsclient -geometry option (X Toolkit) -display option command-line options-display Most X clients accept two powerful and extremely useful options: the -geometry option, which allows you to specify the size and location of a window on the screen; and the -display option, which allows you to specify on which screen a window should be created. (Most commonly, you'd use the -display option to run a client on a remote machine and display the window locally, that is, on your display.) The next few sections illustrate some typical uses of these important options. In explaining how to use them, we introduce some new, perhaps somewhat involved concepts (such as the way distances can be measured on your screen). Bear with us. We feel that you need to master the -geometry and -display options in order to begin to take advantage of the powers of X. After explaining these options in detail, we'll briefly consider some of the characteristics you can specify using other common options. windowsgeometry windowssize and location The command-line option to specify a window's size and location has the form: -geometry geometry The -geometry option can be (and often is) abbreviated as -g, unless the client accepts another option that begins with “g.” The parameter to the geometry option (geometry), referred to as the “standard geometry string,” has four numerical components, two specifying the window's dimensions and two specifying its location. The standard geometry string has the syntax: widthxheight±xoff±yoff Obviously, the first half of the string provides the width and height of the window. Many application windows are measured in pixels. However, application developers are encouraged to use units that are meaningful in terms of the application. For example, xterm's dimensions are measured in columns and rows of font characters. More precisely, an xterm window is some number of characters wide by some number of lines high (80 characters wide by 24 lines high by default). The second half of the geometry string gives the location of the window relative to the horizontal and vertical edges of the screen. Imagining the screen to be a grid (where the upper-left corner is 0,0), xoff (x offset) and yoff (y offset) represent the x and y coordinates at which the window should be displayed. The x and y offsets are measured in pixels. pixels Many users may not be accustomed to thinking in terms of pixels. What exactly is a pixel? A pixel is the smallest element of a display surface that can be addressed by a program. You can think of a pixel as one of the tiny dots that make up a graphic image, such as that displayed by a terminal or a television. The number of dots (or pixels) per inch of screen determines the screen's resolution.The more dots per inch, the higher the resolution (and, hypothetically, the sharper the picture). Hardware manufacturers generally equate resolution with the screen's overall dimensions in pixels. Thus, you also need to know the actual physical size of the monitor (in inches) to determine the dots per inch. We find the dpi figure more useful. There are other factors that determine the “picture quality” of a monitor, including “depth,” or the number of bits per pixel. Depth relates to how many colors a monitor can display. See “How Many Colors are Available on My Screen?”in &cmtf12;, for more information. Since a pixel is a tiny unit of measurement, gauging sizes and distances in pixels will take some practice. For instance, what are the dimensions of your screen in pixels (its resolution)? The xdpyinfo client, described in Chapter 8, will tell you this; your workstation or X terminal documentation may also provide this information. xdpyinfo also tells you the screen's resolution in dots per inch. Keep in mind that monitors can vary substantially. The Sun 19-inch monitor has dimensions of 1152 × 900 pixels and a resolution of 90 × 90 dots per inch (commonly abbreviated to dpi). The NCD 16-inch X terminal has dimensions of 1024 × 1024 pixels and a resolution of 106 × 106 dpi. What are the implications of such hardware differences in specifying client geometry? The size and location of client windows is related to the size and resolution of your screen. For example, if you specify a window with dimensions of 125 × 125 pixels, it will appear somewhat larger on the Sun monitor than on the NCD X terminal. So how do we use the geometry option to specify a window's size and location? First, be aware that you can specify any or all elements of the geometry string. Incomplete geometry specifications are compared to the application's default settings and missing elements are supplied by these values. All client windows have a default size. For example, if you run an xterm window with the geometry option and specify a location but no dimensions, the application default size of 80 characters by 24 lines is used. If you don't specify the x and y coordinates at which to place a client window, the client may provide a default location; if the application doesn't provide a default and mwm is running, the window manager will automatically position the window in the upper-left quarter of the screen. For now, let's just specify a window's location and let the size be the application default. The x and y offsets can be either positive or negative. If you specify positive offsets, you're positioning the left side and top side of the window. Negative offsets are interpreted differently. The possible values for the x and y offsets and their effects are shown in Table 3-1. For example, the command line: % xclock -geometry -10+10 places a clock of the default size in the upper-right corner of the display, 10 pixels from the right and top edges of the screen. To place a window in any of the four corners of the screen, flush against its boundaries, use the following x and y offsets. If you want a window placed away from one or both edges of the screen, the guesswork starts. How many pixels away from the left side of the screen? How many pixels down from the top? You'll have to experiment with placing some clients on your screen to get a better idea of x and y offsets. It's actually a good idea to start some windows and move them around on your screen using the pointer. While you're dragging a window, check the x and y offsets displayed in the small box mwm places in the center of the screen. These coordinates are the positive x and y offsets of the window (i.e., the offsets relative to the upper-left corner of the screen). This method for gauging location is fairly reliable. There seems to be a very slight delay between pointer motion and update to the mwm coordinate box. When you finish dragging the window, the last coordinates visible in the box may differ from the true coordinates by a pixel in either or both directions. But this variance is so trivial that you can supply the coordinates as part of the geometry string and come very close. You can also place some windows in different places by dragging and then determine their geometry specifications using the xwininfo client, described in &cmtf08;. Note that when mwm is running, you should use xwininfo with the -frame option. Now what about the size of a window? For xterm, the size of the window is measured in characters and lines (by default 80 characters by 24 lines). If you want to use a large VT100 window, say 100 characters wide by 30 lines long, you could use this geometry specification: % xterm -geometry 100x30-0-0 This command creates a large xterm window in the lower-right corner of the screen, as illustrated by Figure 3-15.

As stated previously, most of the standard clients (other than xterm) are measured in pixels. For example, xclock is 164 pixels square by default (exclusive of the mwm frame). A client's default dimensions may appear on its reference page in Part Three of this guide. However, you'll probably need to experiment with specifying sizes (as well as locations) on your display. (See &cmtf08;, and the client reference page, for more about xclock.) resource variablesspecifying window size and location Be aware that the geometry option is not necessarily the only means available to specify window size and location. Several clients, including xterm, allow you to set the size and location of a window (and often its icon or an alternate window) using resource variables variablesresourceresource variables (in an .Xresources or other defaults file). We'll introduce some of the basics of specifying resources later in this chapter. See &cmtf11;, for more detailed instructions. See the appropriate client reference pages in Part Three of this guide for a complete list of available resources. You should be aware that, as with all user preferences, you may not always get exactly what you ask for. Clients are designed to work with a window manager, which may have its own rules for window or icon size and placement. However, priority is always given to specific user requests, so you won't often be surprised. windowsgeometry windowssize and location -geometry option (X Toolkit) remote systemrunning client on networkrunning client over clientsdisplay options clientsrunning on another machine clientsoptions command-line options-display -display option DISPLAY environment variable optionscommand-linecommand-line options We have yet to take advantage of X's networking capabilities. Remember that X allows you to run a client on a remote machine across a network. Generally, the results of a client program are displayed on a screen connected to the system where the client is running. However, if you are running a client on a remote system, you probably want to see the results on your own display (connected to a local server). Running a client on a remote machine may give you access to different software, it may increase the efficiency of certain processes, or benefit you in a number of other ways. We discussed some of the advantages of running a client on a remote machine in &cmtf01;. See the section “X Architecture Overview” for details. But how does running a client on a remote system affect how you work with the X display? Once a client is running, it doesn't. You can display the application window on your own screen, enter input using your own keyboard and pointer, and read the client's output in the window on your screen--all while the actual client process occurs on another machine. -display option In order to run a client remotely and display its results locally, you must tell the client process where to display its window. For this purpose, X provides a command-line option: -display. Think of -display as a pointer to the physical display on which you want the window to appear. Like -geometry, the -display option is recognized by most X clients. The display option tells the client on which server to display results (i.e., create its window). The option has the syntax: -display host:server.screen The -display option can be abbreviated as -d, unless the client accepts another option that begins with “d.” The argument to the -display option is a three-component display name. The host specifies the machine on which to create the window, the server specifies the server number, and the screen specifies the screen number. In this context, “host” refers to the Internet address name of the display hardware. It might be the name of a single-user workstation, an X terminal, a PC running an X server, or another hardware device. “Server” refers to an instance of the X server program, which controls a single physical display. An X display may be composed of multiple screens, but the screens share one keyboard and pointer. Most workstations have only one keyboard and pointer and thus are classified as having only one display. Multiuser systems may have multiple independent displays, each running a server program. If one display exists, as in the case of most workstations and X terminals, it is numbered 0; if a machine has several displays, each is assigned a number (beginning with 0) when the X server for that display is started. Similarly, if a single display is composed of multiple screens (sharing one keyboard and pointer), each screen is assigned a number (beginning with 0) when the server for that display is started. Multiple screen displays may be composed of two or more physical monitors. Alternatively, two screens might be defined as different ways of using the same physical monitor. For example, on the Sun-3/60 color workstation, screen 0 is color, and screen 1 is monochrome. Each screen is the size of the monitor; you can only view one screen at a time. In practice, the two screens seem to be side by side: you can “scroll” between them by moving the pointer off either horizontal edge of the screen. Note that the server parameter of the display option always begins with a colon (a double colon after a DECnet node), and that the screen parameter always begins with a period. If the host is omitted or is specified as unix, the local machine is assumed. If the screen is omitted, screen 0 is assumed. By convention, DECnet node names end with a colon. Although much of the current X Window System documentation suggests that any of the parameters to the -display option can be omitted and will default to the local node, server and screen 0, respectively, we have not found this to be true. In our experience, only the host and screen parameters (and the period preceding screen) can be omitted. The colon and server are necessary in all circumstances. DISPLAYenvironment variable Technically speaking, the -display option allows you to override the contents of the DISPLAY environment variable, displayname where stored which stores the display name on UNIX systems. On UNIX systems, the display name is stored in the DISPLAY environment variable. Clients running on the local system access this variable to determine which physical display to connect to (for most clients, “connecting” is equivalent to opening a window). The DISPLAY variable is set automatically by the xterm terminal emulator. Thus it is set when you start X and the first xterm window. For most single-user systems, such as workstations, xterm sets the server and screen numbers to 0, and either omits a hostname (the local host is assumed) or sets the hostname to “unix,” a generic name, which also defaults to the local host. If you are working on a single-user system and run all processes on it, you don't have to deal with issues concerning the display setting. Clients running locally access the DISPLAY variable and open windows on a display connected to the local host. If you are using an X terminal, it should already be configured so that the DISPLAY variable is set properly when you log on to the local host. Again, if you run all process on the local machine (to which your terminal is connected), you don't have to deal with specifying the display. Clients will access the DISPLAY variable and open windows on your X terminal screen. Complications arise when you want to run a process on a remote machine and display the results locally. A client running on a remote machine does not have access to the DISPLAY variable on the local machine. By default, a client running on a remote machine checks the DISPLAY setting on that machine. You can override the DISPLAY environment variable a client accesses by using the -display option when you run the command. -display optionusing Say you're using a single display workstation and the display also has only one screen. The hostname of the workstation is kansas. In order to tell a client to connect to a display, you must identify it by its unique name on the network. (You cannot identify your display by the shorthand setting given to it by xterm--unix:0.0, :0.0 or some variation.) Let's assume that the complete display name for the workstation kansas is: kansas:0.0 Now let's say you want to run an xterm window on a faster system--let's call it oz--on your network. In order to run an xterm on oz but display the window on your screen connected to kansas (the local server), you would run the xterm command using a remote shell (rsh): rsh command % rsh oz xterm -display kansas:0.0 The command to run the remote process might be different depending on the available networking software. Ask your system administrator for the proper command. The xterm process runs on oz, but you've directed the client to use the display and screen numbered 0 on kansas, your local system. Notice that kansas:0.0 is the complete display name. Hypothetically, if the workstation (kansas) has only one screen or it has multiple screens but you want to specify screen 0, you can omit the screen number and the preceding period (.0). Keep in mind that for this process to succeed, the remote system (oz) must have permission to “open” the local display (on kansas). (See “Server Access Control” in &amtfA;, and the xhost reference page in Part Three of this guide, for more information.) If oz has not been granted access to the server running on kansas, the window will not be opened, and you may also get an error message similar to: Can't Open displayerror message error messagesCan't Open display Error: Can't Open display If your command fails, try entering the command: % xhost + in an xterm window running on your display. Then run the remote shell (rsh) again. If problems persist, consult your system administrator or &bmtf08;. The following command illustrates a common mistake: % rsh oz xterm If you try to run a client using a remote shell and forget to direct the client to create its window on your own display, the window will not be displayed and you'll get an error message stating that the display cannot be opened. If you're using an X terminal, it will have a name unique to the terminal (e.g., ncd8), which should be used as the host component of the -display argument. You should not use the name of the system to which the terminal is connected. When you run processes on that machine, you don't have to use -display. If you want to run a client on another machine on the network, you must use -display to point back at your terminal. Say we have a terminal with the full display name: ncd8:0.0 connected to kansas and we want to run an xterm on oz. The command: % rsh oz xterm -display ncd8:0.0 will open the window on our X terminal. In addition to specifying a local display, if permissions allow you can also use the display option to open a window on someone else's display. You might want to display a window on another user's screen for instructional purposes. Multiuser systems can even be set up to allow teachers to display educational material simultaneously to several students, each using an X display of some sort. And, of course, you might want to display a window on a friend's screen, just for the fun of it. (The security problems that allow both innocent pranks like this, as well as more serious breaches, are described in &bmtf08;.) If you're working on kansas and you want to open an xterm window on the first display connected to oz, you could use the command: % xterm -display oz:0.0 Note that you can only open a window on another display if the server running that display permits your server access. (Access must be granted from the remote server, perhaps using xhost.) If oz does not allow kansas access, this command will fail and an error message will indicate that the display cannot be opened. A less than obvious repercussion of using -display to run a remote xterm is that the option sets the DISPLAY variable for the new xterm window--and that DISPLAY setting is passed on to all child processes of the client. Therefore, once you run an xterm on a remote system and correctly specify your own display, you can run any number of clients from that xterm and they will all be displayed on your screen automatically (no -display option is necessary). In one of the examples in the preceding section, we ran an xterm on the remote system oz, specifying the local display kansas:0.0 with the -display option. To query the contents of the DISPLAY variable in the resulting xterm (under the C shell), use the command: % echo $DISPLAY The system should echo: % kansas:0.0 verifying that the display name has been passed to the DISPLAY variable in the new xterm window. You can then run any client you want on oz by entering the command in this xterm window and the window will automatically be displayed on kansas:0.0. The DISPLAY setting will also be passed to any children of this process as well, and will be propagated for any number of “generations.” remote systemlogging in to DISPLAY environment variablesetting after remote login rloginsetting DISPLAY telnetsetting DISPLAY If you log in to a remote UNIX system using rlogin (or telnet) in an xterm window, it's a good idea to set the DISPLAY variable in the new shell to reflect your local display. Then if you run a client process from this window, the new window will be placed on your local display and the DISPLAY setting will be passed on to all child processes. When you set the DISPLAY variable from the command line, the syntax varies depending on the UNIX shell running. The following command sets the variable under the C shell. % setenv DISPLAY kansas:0.0 To set the DISPLAY variable under the Bourne shell, use: $ DISPLAY=kansas:0.0; export DISPLAY DISPLAY environment variable environment variablesDISPLAY variablesenvironmentenvironment variables remote systemmonitoring load on load average A client you may wish to run on another machine is xload, which is used to keep track of the system load average. xload (poll system load average) By default, xload polls the system for the load average at ten-second intervals and displays the results in a simple histogram. If you are running processes on more than one machine, it's useful to gauge the level of activity on the systems in question. This information should help you judge when to start processes and monitor how your processes are impacting system resources. Say you're running clients both on the local machine kansas and on the remote machine oz. On the local display, you can have two xload windows, one showing activity on kansas and another showing activity on oz. To create an xload window monitoring activity on kansas, use the command: % xload Once the xload window is created, move it to a convenient location on the screen. Then run an xload process on oz using a remote shell and display the results in a window on kansas: % rsh oz xload -display kansas:0.0 The display option tells xload to create its window on the local display (kansas). Again, move the window using the pointer. Figure 3-16 shows the resulting kansas display: two xload windows--the top window monitoring activity on the local system and the bottom one monitoring activity on the remote system.

clientsstarting additional Now that we've learned something about the tools of the display, how to size and position windows, and run remote processes, let's try to set up a useful working display. Say we're using a Sun 3/60 workstation with the hostname jersey. The workstation has a single display with two screens: screen 0 is color and screen 1 is black and white. Once X and the window manager are running, we might set up the display using the following commands. Note that you should run mwm with the -multiscreens option to have the program manage all screens on the display. If you start mwm without this option, it only manages screen 0. In this case, you can either kill and restart it with -multiscreens or run another instance of it on screen 1. First, run an xterm on a more powerful remote system called manhattan and place it on screen 0 of jersey. % rsh manhattan xterm -geometry +0-0 -display jersey:0.0 Run xload windows on both jersey and manhattan to monitor loads on these systems. Again, place the windows on jersey's color screen in convenient locations. % xload -geometry -10-200 % rsh manhattan xload -geometry -10-20 -display jersey:0.0 Run another xterm window on jersey. % xterm -geometry +50-0 Then iconify the login xterm window so that you don't inadvertently kill it (and shut down X in the bargain). Remember: to iconify a window place the pointer on the Minimize command button on the window's frame and click the first button. (See “Converting a Window to an Icon” earlier in this chapter.) Run an oclock. % oclock -geometry +0+0 Figure 3-17 shows the jersey display, screen 0: a fairly useful layout.

You might also place client windows on the workstation's alternate screen, the black and white screen numbered 1. By default, windows are always placed on screen 0 but you can place a client window on screen 1 by specifying the screen number in the -display option when starting the client. For instance, each of the following commands places an xterm window on screen 1. % xterm -display jersey:0.1 % rsh manhattan xterm -geometry -0-0 -display jersey:0.1 Figure 3-18 illustrates screen 1.

As we'll see in &amtfA;, on most systems you can place the commands you run to set up your display in a special file that is invoked when you log in. Once this file (usually called either .xinitrc or .xsession) is in place, when you log in your display will be set up to your specifications automatically. Notice that the commands we used to set up jersey illustrate the power of the -geometry and -display options to create a working environment that suits individual needs. However, these options barely hint at the number of features you can specify for each client. The following section introduces some principles of client customization. Part Two of this guide examines customization in depth. command-line options In a sense, command-line options allow you to customize one program. We've already seen how to use the -geometry and -display options, which are accepted by most clients. &cmtf10;, describes some of the other options accepted by most of the standard clients. These options set window features such as: The font in which text is displayed. The background color. The foreground color (such as the color of text). The text displayed in the title area. The text of an icon label. Many clients also accept a large number of application-specific options (listed on the reference page for each client in Part Three of this guide). Using a combination of standard and application-specific options, you can tailor a client to look and behave in ways that better suit your needs. Like most clients, xclock accepts a variety of options. Some of xclock's options are intended to enhance the clock display aesthetically and some to affect its operation. Taking a look at a few of xclock's options should give you a better idea of the flexibility of X. The following command line runs a custom xclock display: % xclock -hd green -hl royalblue -bg lightblue -fg royalblue -update 1 -chime -hd option command-line options-hd -hl option command-line options-hl -bg option (X Toolkit) command-line options-bg (background) command-line options-fg -fg option (X Toolkit) -update option command-line options-update As you can see, these specifications are intended for a color monitor. (X is highly flexible in the use of color. See &cmtf12;, for more information.) The -hd option sets the color of the clock's hands to green. The -hl option provides even more detail, specifying the color of the outline of the hands as royal blue. -bg and -fg are two of the options accepted by most clients; they set the window's background and foreground color, in this case to light blue and royal blue, respectively. The -update option takes as its argument the frequency in seconds at which the time on the clock should be updated. We've specified that the time be updated at one second intervals. Thus a second hand (also green) will be added to the xclock display. (The xclock reference page in Part Three of this guide specifies that a second hand is added if -update is given an argument of less than 30 seconds.) -chime option The -chime option specifies that the keyboard bell will ring once on the half hour and twice on the hour. These options create a somewhat fancy xclock display. You might or might not want to use so many options, but these and several more are available. xclock (analog or digital clock) By default, xclock displays a traditional clock face (an analog clock). You can create a digital xclock using the following option: % xclock -digital The digital xclock is pictured in Figure 3-19.

Logically, the -hd and -hl options, which set the color of the clock hands, are only valid with the analog (default) xclock. For a complete list of options, see the xclock reference page in Part Three of this guide. Command-line options override the default characteristics of a client for the single client process. Traditional UNIX applications rely on command-line options to allow users to customize the way they work. X also offers many command-line options, but these options have some limitations and liabilities. First, the number of client features that can be controlled by command line options is limited. Most applications have many more customizable features than their command-line options indicate. Actually, a client can have so many customizable features that typing a command line to set them all would be impractical. And if you generally use the same options with a client, it is tedious (and a waste of time) to type the options each time you run the program. clientsspecifying default characteristics for X offers an alternative to customizing a single client process on the command line. You can specify default characteristics for a client using variables called resources. command-line options clientscustomizing customizingclients X Window Systemenvironment customizing customizingX environment Command-line options allow you to customize one instance of a client program. In addition, X provides a mechanism that allows you to specify characteristics that take effect every time you run a client. Almost every feature of a client program can be controlled using a variable called a resource. You can change the behavior or appearance of a program by changing the value associated with a resource variable. (In some cases, a resource variable controls the same characteristic as a command-line option. However, while the option specifies a characteristic for the single client process being invoked, a resource variable makes the characteristic the program default.) resource variablesfile kept in .Xdefaults file .Xresources file You generally place resource specifications in a file in your home directory. (The file can have any name, but is often called .Xresources or .Xdefaults.) The resources you specify are one of several factors that affect the appearance and behavior of a client. By default, the way a client looks and behaves is determined by the program code, and in some cases, by a system-wide file of application defaults. Several clients have application defaults files that determine certain client features. For xterm, the application defaults specify such things as the labels for menu items, the fonts used to display menu items, and the shape of the pointer when it's in an xterm window. Application defaults files generally reside in the directory /usr/lib/X11/app-defaults and are named for the client application. In describing the appearance and behavior of clients in this guide, we assume all of the standard application defaults files are present on your system and accessible by the client programs. If, by some chance, a client's application defaults file has been edited or removed from your system, the client may not look or behave exactly as we describe it. If a client application appears substantially different than depicted in this guide, you may be using a different version of the program or the application defaults may be different. Consult your system administrator. Within an application defaults file, defaults are set using resources. The resources specified in a client's application defaults files are usually just a subset of a greater number of resources that can be set. If the characteristics you set in your own resources file already have system-wide application defaults, your own settings take precedence. resourcessetting Keep in mind, however, that command-line options override both your own defaults and any system-wide defaults for the single client process. To make your resource specifications available to all clients, X provides a program called xrdb, the X resource database manager. xrdb stores resources directly in the server where they are accessible to all clients, regardless of the machine the clients are running on. The basic syntax of a resource specification is fairly simple. Each client recognizes certain resource variables that can be assigned a value. The variables for each client are listed on its reference page in Part Three of this guide. A resource definition file is basically a two-column list, where each line specifies a different resource. The simplest resource definition line has the name of the client, followed by an asterisk, and the name of the variable, followed by a colon, in the left column. The right column (separated from the left by a tab or whitespace) contains the value of the resource variable. client*variable:value The following example shows five simple resource specifications for the xclock client. These particular resources specify the same characteristics as the command-line options we used to create the green and blue xclock in the preceding section. xclock*hands: green xclock*highlight: royalblue xclock*background: lightblue xclock*foreground: royalblue xclock*update: 1 xclock*chime: true To set up your environment so that these characteristics apply each time you run xclock, you would perform the following steps: In your home directory, create a file containing the resources listed in Example 3-1. Name the file .Xresources. (A resource file can actually have any name, but is often called .Xresources or .Xdefaults.) Load the resources into the server by entering the following command in an xterm window: % xrdb -load .Xresources Then each time you run xclock without options (for the remainder of that login session), the window will reflect the new defaults. You should load resources using xrdb every time you log in. In &amtfA;, we'll describe how to automate this process using a special startup script, which also opens the client windows you want on your display. If you want to run an application with different characteristics (colors, update frequency, etc.) from the defaults, use the appropriate command line options to override the resource specifications. Resource specifications can be much more complicated than our samples suggest. For applications written with a toolkit (such as the X Toolkit or the Open Software Foundation's Motif Toolkit), X allows you to specify different characteristics for individual components, or widgets, within the application. Typical widgets create graphical features such as menus, command buttons, dialog boxes, and scrollbars. Within most toolkit applications is a fairly complex widget hierarchy--widgets exist within widgets (e.g., a command button within a dialog box). Resource naming syntax can parallel the widget hierarchy within an application. For instance, you might set different background colors for different command buttons and specify still another background color for the dialog box that encloses them. In such cases, the actual widget names are used within the resource specification. &cmtf11;, explains the resource naming syntax in greater detail and outlines the rules governing the precedence of resources. It also explains how to use the editres program to examine a (standard) client's widget hierarchy and set resources accordingly. X Window Systemenvironment customizing customizingX environment resourcessetting clientscustomizing customizingclients There are many useful client programs supplied with the X Window System. Details of how to use one of the most important clients, the xterm terminal emulator, are provided in Chapter 5. Clients to list and display fonts are described in &cmtf06;. Chapter 6 also describes the X font naming conventions and various ways to specify fonts on the command line (and in resource files). Chapter 7 describes several Graphics Utilities available with X. An overview and tutorial for other standard clients and instructions on using certain public domain clients are provided in &cmtf08;. &cmtf09;, gives instructions on using Motif applications. All clients are described in detail in a reference page format in Part Three of this guide. We've introduced some basic operations you can perform using the mwm window manager. For instructions on performing additional window manager operations, such as lowering a window, read &cmtf04;. You can then go on to read more about xterm in Chapter 5 and about some of the other standard clients in Chapters 6 through 8.