Table of Contents

NAME

SYNOPSIS

DESCRIPTION

is an implementation of Kernighan and Bentley’s language for typesetting graphs, as described in ‘‘Grap-A Language for Typesetting Graphs, Tutorial and User Manual,’’ by Jon L. Bentley and Brian W. Kernighan, revised May 1991, which is the primary source for information on how to use As of this writing, it is available electronically at Additional documentation and examples, packaged with may have been installed locally as well. If available, paths to them can be displayed using or (or / This version is a black box implementation of and some inconsistencies are to be expected. The remainder of this manual page will briefly outline the language as implemented here. is a pre-processor. It takes commands embedded in a source file which are surrounded by and macros, and rewrites them into commands to display the graph. Other lines are copied. Output is always to the standard output, which is usually redirected. Input is from the given which are read in order. A of is the standard input. If no are given, input is read from the standard input. Because is a preprocessor, and GNU will output TeX, it is possible to use with TeX. The option specifies a file of macro definitions to be read at startup, and defaults to /usr/local/share/grap/grap.defines . The option inhibits the reading of any initial macros file (the flag is a synonym for though I do not remember why). The defines file can also be given using the environment variable. (See below). prints the version information on the standard output and exits. is a synonym for makes labels unaligned by default. This version of uses new features of GNU to align the left and right labels with the axes, that is that the left and right labels run at right angles to the text of the paper. This may be useful in porting old programs. makes plot strings unclipped by default. Some versions of allow users to place a string anywhere in the coordinate space, rather than only in the frame. By default this version of does not plot any string centered outside the frame. allows strings to be placed anywhere. See also the and string modifiers described in the statement. is followed by a colon-separated list of directories used to search for relative pathnames included via The path is also used to locate the defines file, so if the changes the defines file name to a relative name, it will be searched for in the path given by The search path always includes the current directory, and by default that directory is searched last. All numbers used internally by are double precision floating point values. Sometimes using floating point numbers has unintended consequences. To help avoid these problems, can use two thresholds for comparison of floating point numbers, set by or The flag sets coarse comparison mode, which is suitable for most applications. If you are plotting small values en less than 1e-6 or so en consider using which uses very fine comparisons between numbers. You may also want to rescale your plotted values to be larger in magnitude. The coarse comarisons are used by default. To be precise, the value by which two numbers must differ for to consider them not equal is called the comparison limit and the smallest non-zero number is called the minimum value. The values a given version of uses for these are included in the output of or All commands are included between and macros, which are consumed by The output contains between and macros. Any arguments to the macro in the input are arguments to the macro in the output, so graphs can be scaled just like diagrams. If is given, any macro beginning with .G1 or .G2 is treated as a .G1 or .G2 macro, for compatibility with old versions of troff. Using also forces pure troff syntax on embedded font change commands when strings have the attribute, and all strings to be The flag prints a brief help message and exits. is a synonym for It is possible for someone to cause to fail by passing a bad format string and data to the command. If is integrated as part of the printing system, this could conceivably provided a path to breaching security on the machine. If you choose to use as part of a printing system run by the super-user, you should disable commands. This can be done by calling with the flag, setting the environment variable, or compiling with the GRAP_SAFER preprocessor symbol defined. (The GNU configure script included with will define that preprocessor symbol if the option is given.) The commands are sketched below. Refer to Kernighan and Bentley’s paper for the details. New versions of will invoke if is given. Commands are separated from one another by newlines or semicolons (;). ...

... This describes how the axes for the graph are drawn. A is a line description, e.g., or the literal It may also include a keyword followed by the color to draw the string in double quotes. Any color understood by the underlying groff system can be used. Color can only be used under GNU pic, and is not available in compatibility mode. Similarly, for pic implementations that understand that attribute may be used with a real valued parameter. is not available in compatibility mode. If the first is given, the frame is drawn with that style. The default is The height and width of the frame can also be specified in inches. The default line style can be over-ridden for sides of the frame by specifying additional parameters to If no plotting commands have been given before the command is issued, the frame will be output at that point in the plotting stream relative to embedded or commands. Otherwise the frame is output before the first plotted object (even invisible ones). and are in inches by default, but can be any unit. If omitted, the dimensions are 2 inches high by 3 inches wide. The command specifies a new coordinate system or sets limits on the default system. It defines the largest and smallest values that can be plotted, and therefore the scale of the data in the frame. The limits for the x and y coordinate systems can be given separately. If a is given, that coordinate system is defined, if not the default system is modified. A coordinate system created by one command may be modified by subsequent commands. A program may declare a coordinate space using a file of data through a macro that plots the data and finds its maxima and minima, and then define the size of the coordinate system with a second statement. This command also determines if a scale is plotted logarithmically. means the same thing as The command defines the style with which a given line will be plotted. If is given, the style is associated with that name, otherwise the default style is set. is a line description, and the optional is a string to be centered at each point. The default line description is and the default plotting string is a centered bullet, so by default each point is a filled circle, and they are unconnected. If points are being connected, each command ends any current line and begins a new one. When defining a line style, that is the first command for a given line name, specifying no plot string means that there are to be no plot strings. Omitting the plot string on subsequent commands addressing the same named line means not to change the plot string. If a line has been defined with a plot string, and the format is changed by a subsequent statement, the plot string can be removed by specifying "" in the statement. The plot string can have its format changed through several string_modifiers. String_modifiers are described in the description of the command. The standard defines file includes several macros useful as plot strings, including and is a synonym for The command plots the given point using the line style given by or the default if none is given. If is given, it should have been defined by an earlier command, if not a new line style with that name is created, initialized the same way as the default style. The two expressions give the point’s x and y values, relative to the optional coordinate system. That system should have been defined by an earlier command, if not, grap will exit. If the optional is given, it overrides the style’s default line description. You cannot over-ride the plotting string. To use a different plotting string use the command. The coordinates may optionally be enclosed in parentheses: ... These commands both plot a string at the given point. In the first case the literal strings are stacked above each other. The string_modifiers include the justification modifiers and and absolute and relative modifiers. See the documentation for the description of the justification modifiers. also supports the and modifiers which are briefly noted in the description of the command. The standard defines file includes several macros useful as plot strings, including and Strings placed by either format of the command are restricted to being within the frame. This can be overriden by using the attribute, which allows a string to be plotted in or out of the frame. The and flags set on all strings, and to prevent a string from being plotted outside the frame when those flags are active, the attribute can be used to retore clipping behavior. Though or can be applied to any string, it only has meaning for statements. sets the string size to points. If is preceded by a + or -, the size is increased or decreased by that many points. If and a color name in double quotes appears, the string will be rendered in that color under a version of GNU troff that supports color. Color is not available in compatibility mode. In the second version, the is converted to a string and placed on the graph. is a format string. Only formatting escapes for printing floating point numbers make sense. The format string is only respected if the command is also active. See the description of for the various ways to disable it. and respond differently when is running safely. ignores any arguments, passing the format string through without substitution. ignores the format string completely, plotting using the format. Points are specified the same way as for commands, with the same consequences for undefined coordinate systems. The second form of this command is because the first form can be used with a expression (See This command controls the placement of ticks on the frame. By default, ticks are automatically generated on the left and bottom sides of the frame. The first version of this command turns on the automatic tick generation for a given side. The or parameter controls the direction and length of the ticks. If a is specified, the ticks are automatically generated using that coordinate system. If no system is specified, the default coordinate system is used. As with and the coordinate system must be declared before the statement that references it. This syntax for requesting automatically generated ticks is an extension, and will not port to older implementations. The second version of the command overrides the automatic placement of the ticks by specifying a list of coordinates at which to place the ticks. If the ticks are not defined with respect to the default coordinate system, the parameter must be given. For each tick a style format string can be given. The defaults to The format string can also take string modifiers as described in the command. To place ticks with no labels, specify as If is disabled, behaves as with respect to the format string. The labels on the ticks may be shifted by specifying a direction and the distance in inches to offset the label. That is the optional direction and expression immediately preceding the The third format of the command over-rides the default tick generation with a set of ticks ar regular intervals. The syntax is reminiscent of programming language for loops. Ticks are placed starting at ending at one unit apart. If the clause is specified, ticks are units apart. If an operator appears before each tick is operated on by that operator instead of +. For example ticks left out from 2 to 32 by *2 will put ticks at 2, 4, 8, 16, and 32. If is specified, all ticks are formatted using it. The parameters preceding the act as described above. The and forms of tick command may both be issued on the same side of a frame. For example: ticks left out from 2 to 32 by *2 ticks left in 3, 5, 7 will put ticks on the left side of the frame pointing out at 2, 4, 8, 16, and 32 and in at 3, 5, and 7. The final form of turns off ticks on a given side. If no side is given the ticks for all sides are cancelled. is a synonym for The command is similar to the command except that specifies the placement of lines in the frame. The syntax is similar to as well. By specifying in the command, no ticks are drawn on that side of the frame. If ticks appear on a side by default, or have been declared by an earlier command, does not cancel them unless is specified. Instead of a direction for ticks, allows the user to pick a line description for the grid lines. The usual line descriptions are allowed. Grids are labelled by default. To omit labels, specify the format string as If is disabled, behaves as with respect to the format string. ... The command places a label on the given axis. It is possible to specify several labels, which will be stacked over each other as in The final argument, if present, specifies how many inches the label is shifted from the axis. By default the labels on the left and right labels run parallel to the frame. You can cancel this by specifying as a This draws an circle at the point indicated. By default, the circle is small, 0.025 inches. This can be over-ridden by specifying a radius. The coordinates of the point are relative to the named coordinate system, or the default system if none is specified. This command has been extended to take a line description, e.g., It also accepts the filling extensions described below in the command. It will also accept a keyword that gives the color of the outline of the circle in double quotes and a command that sets the color to fill the circle with similarly. Colors are only available when compatibility mode is off, and using a version of GNU pic that supports color. This draws a line or arrow from the first point to the second using the given style. The default line style is The can be given either before the or after the clause. If both are given the second is used. It is possible to specify one point in one coordinate system and one in another, note that if both points are in a named coordinate system (even if they are in the same named coordinate system), both points must have given. The command imports data from another file into the current graph. The form with only a filename given is a simple file inclusion; the included file is simply read into the input stream and can contain arbitrary commands. The more common case is that it is a number list; see below. The second form takes lines from the file, splits them into words delimited by one or more spaces, and calls the given macro with those words as parameters. The macro may either be defined here, or be a macro defined earlier. See for more information on macros. The may be omitted if the clause is present. If so the current file is treated as the input file until is encountered at the beginning of the line. is one of the workhorses of Check out the paper and for more details. Confirm the location of the examples directory using the flag. Prints its argument to the standard error. This passes to Unlike K&B no macro or variable expansion is done. I believe that this is also true for GNU version 1.10. See the section for information on defining blocks. This issues the given statements in the enclosing and at the point where the command is issued. Statements that begin with a period are considered to be and are output in the enclosing and at the point where the command appears. For the purposes of relative placement of or commands, the frame is output immediately before the first plotted object, or the statement, if any. If the user specifies or commands and neither any plotable object nor a command, the commands will not be output. This command is used to position graphs with respect to each other. The current graph is given the name (names used by begin with capital letters). Any commands following the graph are used to position the next graph. The frame of the graph is available for use with name The following places a second graph below the first: graph Linear [ graph description ] graph Exponential with .Frame.n at \    Linear.Frame.s - (0, .05)
[ graph description ] This assigns to the variable has only numeric (double) variables. Assignment creates a variable if it does not exist. Variables persist across graphs. Assignments can cascade; assigns 35 to and The command facilitates drawing bar graphs. The first form of the command describes the bar somewhat generally and has place it. The bar may extend up or to the right, is centered on and extends up or right units (in the given coordinate system). For example bar up 3 ht 2 draws a 2 unit high bar sitting on the x axis, centered on x=3. By default bars are 1 unit wide, but this can be changed with the keyword. By default bars sit on the base axis, i.e., bars directed up will extend from y=0. That may be overridden by the keyword. (The bar described above has corners (2.5, 0) and (3.5, 2).) The line description has been extended to include a keyword that specifies the shading inside the bar. Bars may be drawn in any line style. They support the and keywords described under The second form of the command draws a box with the two points as corners. This can be used to draw boxes highlighting certain data as well as bar graphs. Note that filled bars will cover data drawn under them. The statement provides simple conditional execution. If is non-zero, the after the statement is executed. If not the after the is executed, if present. See for the definition of blocks. Early versions of this implementation of treated the blocks as macros that were defined and expanded in place. This led to unnecessary confusion because explicit separators were sometimes called for. Now, inserts a separator (;) after the last character in so constructs like if (x == 3) { y = y + 1 } x = x + 1

behave as expected. A separator is also appended to the end of a block. This command executes iteratively. The variable is set to and incremented by until it exceeds The iteration has the semantics defined in the command. The definition of is discussed in See also the note about implicit separators in the description of the command. An can be used in place of supports most standard arithmetic operators: + - / * ^. The carat (^) is exponentiation. In an statement also supports the C logical operators ==, !=, &&, || and unary !. Also in an == and != are overloaded for the comparison of quoted strings. Parentheses are used for grouping. Assignment is not allowed in an expression in any context, except for simple cascading of assignments. works as expected; does not execute. supports the following functions that take one argument: The logarithms are base 10 and the trigonometric functions are in radians. returns Euler’s number to the given power and returns the natural logarithm. The natural log and exponentiation functions are extensions and are probably not available in other implementations. returns a random number uniformly distributed on [0,1). The following two-argument functions are supported: works just like The random number generator can be seeded by calling with a single parameter (converted internally to an integer). Because its return value is of no use, you must use as a separate statement, it is not part of a valid expression. is not portable. The function takes no arguments and returns the process id. This may be used to seed the random number generator, but do not expect cryptographically random values to result. Other than string comparison, no expressions can use strings. One string valued function exists: ). It operates like except returning the value. It can be used anywhere a quoted string is used. If is run with the environment variable is defined, or has been compiled for safer operation, the command will return the format string. This mode of operation is only intended to be used only if is being used as part of a super-user enabled print system. has a simple but powerful macro facility. Macros are defined using the command :
Every occurrence of in the program text is replaced by the contents of is defined by a series of statements in nested { }’s, or a series of statements surrounded by the same letter. An example of the latter is define foo X coord x 1,3 X Each time appears in the text, it will be replaced by Macros are literal, and can contain newlines. If a macro does not span multiple lines, it should end in a semicolon to avoid parsing errors. Macros can take parameters, too. If a macro call is followed by a parenthesized, comma-separated list the values starting with $1 will be replaced in the macro with the elements of the list. A $ not followed by a digit is left unchanged. This parsing is very rudimentary; no nesting or parentheses or escaping of commas is allowed. Also, there is no way to say argument 1 followed by a digit (${1}0 in sh(1) ). The following will draw a line with slope 1. define foo { next at $1, $2 } for i from 1 to 5 { foo(i,i) } Macros persist across graphs. The file contains simple macros for plotting common characters. The command deletes a macro. See the directory for more examples of macros. Confirm the location of the examples directory using the flag. A whitespace-separated list of numbers is treated specially. The list is taken to be points to be plotted using the default line style on the default coordinate system. If more than two numbers are given, the extra numbers are taken to be additional y values to plot at the first x value. Number lists in DWB can be comma-separated, and this supports that as well. More precisely, numbers in number lists can be separated by either whitespace, commas, or both. 1 2 3 4 5 6

Will plot points using the default line style at (1,2), (1,3),(4,5) and (4,6). A simple way to plot a set of numbers in a file named is: .G1 copy "./data" .G2 defines pic macros that can be used in embedded pic code to place elements in the graph. The macros are and These macros define pic distances that correspond to the given argument. They can be used to size boxes or to plot pic constructs on the graph. To place a given construct on the graph, you should add Frame.Origin to it. Other coordinate spaces can be used by replacing with the name of the coordinate space. A coordinate space named cannot be reliably accessed by these macros. The macros are emitted immediately before the frame is drawn. DWB may use these as part of its implementation. This provides them only for compatibility. Note that these are very simple macros, and may not do what you expect under complex conditions.

ENVIRONMENT VARIABLES

If the environment variable is defined, will look for its defines file there. If that value is a relative path name the path specified in the option will be searched for it. overrides the compiled in location of the defines file, but may be overridden by the or flags. If is set, is disabled to prevent forcing to core dump or smash the stack.

FILES

SEE ALSO

If documentation and examples have been installed, or will display the locations.

BUGS

There are several small incompatibilities with K&R They include the command not expanding variables and macros, and a more strict adherence to parameter order in the internal commands. Although much improved, the error reporting code can still be confused. Notably, an error in a macro is not detected until the macro is used, and it produces unusual output in the error message. Iterating many times over a macro with no newlines can run out of memory.

AUTHOR

This implementation was done by contributed many bug fixes, including a considerable revamp of the error reporting code. If you can actually find an error in your code, you can probably thank him. was designed and specified by and


Table of Contents