Tcl/Tk 8.5 Programming Cookbook

So, you've installed Tcl, written some scripts, and now you're ready to get a deeper understanding of Tcl and all that it has to offer. So, why are we starting with the shell when it is the most basic tool in the Tcl toolbox?

When I started using Tcl I needed to rapidly deliver a Graphical User Interface (GUI) to display a video from the IP-based network cameras. The solution had to run on Windows and Linux and it could not be browser-based due to the end user's security concerns. The client needed it quickly and our sales team had, as usual, committed to a delivery date without speaking to the developer in advance. So, with the requirement document in hand, I researched the open source tools available at the time and Tcl/Tk was the only language that met the challenge.

The original solution quickly evolved into a full-featured IP Video Security system with the ability to record and display historic video as well as providing the ability to attach to live video feeds from the cameras. Next search capabilities were added to review the stored video and a method to navigate to specific dates and times. The final version included configuring advanced recording settings such as resolution, color levels, frame rate, and variable speed playback. All was accomplished with Tcl.

Due to the time constraints, I was not able get a full appreciation of the capabilities of the shell. I saw it as a basic tool to interact with the interpreter to run commands and access the file system. When I had the time, I returned to the shell and realized just how valuable a tool it is and the many capabilities I had failed to make use of.

When used to its fullest, the shell provides much more that an interface to the Tcl interpreter, especially in the early stages of the development process. Need to isolate and test a procedure in a program? Need a quick debugging tool? Need real-time notification of the values stored in a variable? The Tcl shell is the place to go.

Since then, I have learned countless uses for the shell that would not only have sped up the development process, but also saved me several headaches in debugging the GUI and video collection. I relied on numerous dialog boxes to pop up values or turned to writing debugging information to error logs. While this was an excellent way to get what I needed, I could have minimized the overhead in terms of coding by simply relying on the shell to display the desired information in the early stages.

While dialog windows and error logs are irreplaceable, I now add in quick debugging by using the commands the shell has to offer. If something isn't proceeding as expected, I drop in a command to write to standard out and voila! I have my answer. The shell continues to provide me with a reliable method to isolate issues with a minimum investment of time.

The Tcl Shell (Tclsh) provides an interface to the Tcl interpreter that accepts commands from both standard input and text files. Much like the Windows Command Line or Linux Terminal, the Tcl shell allows a developer to rapidly invoke a command and observe the return value or error messages in standard output. The shell differs based on the Operating System in use. For the Unix/Linux systems, this is the standard terminal console; while on a Windows system, the shell is launched separately via an executable.

If invoked with no arguments, the shell interface runs interactively, accepting commands from the native command line. The input line is demarked with a percent sign (%) with the prompt located at the start position. If the shell is invoked from the command line (Windows DOS or Unix/Linux terminal) and arguments are passed, the interpreter will accept the first as the filename to be read. Any additional arguments are processed as variables. The shell will run until the exit command is invoked or until it has reached the end of the text file.

When invoked with arguments, the shell sets several Tcl variables that may be accessed within your program, much like the C family of languages. These variables are:

The following recipe illustrates a basic command invocation. In this example, we will use the puts command to output a "Hello World" message to the console.

% puts "Hello World"
Hello World

The expr command is used to evaluate mathematical expressions. This command can address everything from simple addition and subtraction to advanced computations, such as sine and cosine. This eliminates the need to make system calls to perform advanced mathematical functions. The expr command evaluates the input and arguments, and returns an integer or floating-point value.

A Tcl expression consists of a combination of operators, operands, and parenthetical containers (parenthesis, braces, or brackets). There are no strict typing requirements, so any white space is stripped by the command automatically. Tcl supports non-numeric and string comparisons as well as Tcl-specific operators.

Tcl operands are treated as integers, where feasible. They may be specified as decimal, binary (first two characters must be 0b), hexadecimal (first two characters must be 0x), or octal (first two characters must be 0o). Care should be taken when passing integers with a leading 0, for example 08, as the interpreter would evaluate 08 as an illegal octal value. If no integer formats are included, the command will evaluate the operand as a floating-point numeric value. For scientific notations, the character e (or E) is inserted as appropriate. If no numeric interpretation is feasible, the value will be evaluated as a string. In this case, the value must be enclosed within double quotes or braces. Please note that not all operands are accepted by all operators. To avoid inadvertent variable substitution, it is always best to enclose the operands within braces. For example, take a look at the following:

Operands may be presented in any of the following:

Tcl supports a subset of the C programming language math operators and treats them in the same manner and precedence. If a named function (such as sine) is encountered, expr automatically makes a call to the mathfunc namespace to minimize the syntax required to obtain the value.

Tcl expr operators may be specified as noted in the following table, in the descending order of precedence:

Mathematical functions (such as sine and cosine) are replaced with a call to the Tcl mathfunc namespace. This does not require any additional syntax to access the namespace as it is called automatically. These are invoked by passing the Function followed by the value or values to evaluate to the expr command. Those functions that accept multiple arguments require that the arguments be comma delimited. The default Mathematical functions are listed below in alphabetical order. These functions require a specific syntax (for example expr {function(value,value)}) to be accessed, as described in the Computing mathematical expressions section that follows:

In the following examples, we will see the correct syntax for both simple and complex mathematical formulas. To accomplish these computations, we will be using the Tcl expr command. The expr command, as its name implies, is used to evaluate mathematical expressions. This command can address everything from simple addition and subtraction to advanced computations such as sine and cosine. This removes the need to make system calls to perform advanced mathematical functions. The expr command evaluates the input and arguments and returns an integer, floating-point, or string value as appropriate.

A Tcl expression consists of a combination of operators, operands, and parenthetical containers (parenthesis, braces, or brackets). There are no strict typing requirements so any white space is stripped by the command automatically. Tcl supports non-numeric and string comparisons as well as Tcl specific operators.

As you will see, some computations may be performed without parenthetical notations; however, it is best to get into the habit of always using them. For example, expr 1+1 and expr (1+1) will both return a value of 2. While the omission of the parenthetical notation is completely acceptable in this usage of the expr command, I recommend developing the habit of always using them.

My personal favorite is the if-then-else expression. It provides a rapid method for comparison in a "single line" format. For example, if x and y are equal to 10, while z = 4 would be entered as expr ($x?$y:$z). This expression evaluates $x as a Boolean expression. If it's true the expression will return $y; if it's false, it returns $z.

Parenthetical notation is required for any operation that will access a specific mathematical function. For example: expr {pow (8, 4)} will access the mathematical power function and return a value of 4096.

Variable substitution is handled using the Tcl $ notation. The following example uses an x variable with a value of 4 and is entered as expr {pow (8, $x)}. This expression returns a value of 4096 as observed in the previous example. In the second case, $x has been processed with its variable value of 4.

Tcl commands that accept filenames as arguments require that they be in one of three formats, depending on the platform in use. The platform in use is stored in the global TCL_platform array variable, created at the start of the program. Please note that to address issues of portability, you must manually manipulate the formats to ensure that they are annotated correctly.

These formats are absolute, relative, and volume-related.

The following conventions are platform-specific annotations for both the directory structure and the specific filenames.

As with all the programming languages, it is the variable that allows for true flexibility and usability. Tcl differs from some scripted languages, as, there is no need to implicitly declare the variable type. For example a variable of "3" will be stored within Tcl with the same internal representation, as if it have been defined as the integer 3. If the variable is then used in a calculation, Tcl will then convert it to an integer for computation. This is referred to as shimmering in Tcl.

% set x 3
3

% incr x 3
6

% unset x
%

% unset y
can't unset "y": no such variable

% unset nocomplain y
%

With any scripting language, the ability to provide arguments allows you to write a script that accepts arguments to perform a specific function.

As previously discussed, Tcl has several global variables to allow for the passing of command line arguments. The number of command line arguments to a Tcl script is passed as the global variable argc. The name of a Tcl script is passed to the script as the global variable argv0, and the arguments are passed as a list in the argv global variable.

# If no command line arguments are passed perform no actions
if {$argc > 0} {
# Print out the filename of the script
puts "The name of the script is: $argv0"
# Print out the count of the arguments passed
puts "Total count of arguments passed is: $argc"
# Print out a list of the arguments
puts "The arguments passed are: $argv"
# Using the List Index of argv print a specific argument
puts "The first argument passed was [lindex $argv 0]"
}

% Tclsh85 args.Tcl ONE 2 3
The name of the script is: args.Tcl
Total count of arguments passed is: 3
The arguments passed are: ONE 2 3
The first argument passed was ONE
%

% Tclsh85 args.Tcl \home \etc
The name of the script is: args.Tcl
Total count of arguments passed is: 2
The arguments passed are: home etc
The first argument passed was home
%

% Tclsh85 args.Tcl \\home \\etc
The name of the script is: args.Tcl
Total count of arguments passed is: 2
The arguments passed are: {\home} {\etc}
The first argument passed was \home
%