DESCRIPTION

The following global variables are created and managed automatically by the Tcl library. Except where noted below, these variables should normally be treated as read-only by application-specific code and by users.

auto_path

If set, then it must contain a valid Tcl list giving directories to search during auto-load operations (including for package index files when using the default package unknown handler). This variable is initialized during startup to contain, in order: the directories listed in the TCLLIBPATH environment variable, the directory named by the tcl_library global variable, the parent directory of tcl_library, the directories listed in the tcl_pkgPath variable. Additional locations to look for files and package indices should normally be added to this variable using lappend.

Additional variables relating to package management exist. More details are listed in the VARIABLES section of the library manual page.

env

This variable is maintained by Tcl as an array whose elements are the environment variables for the process. Reading an element will return the value of the corresponding environment variable. Setting an element of the array will modify the corresponding environment variable or create a new one if it does not already exist. Unsetting an element of env will remove the corresponding environment variable. Changes to the env array will affect the environment passed to children by commands like exec. If the entire env array is unset then Tcl will stop monitoring env accesses and will not update environment variables.

Under Windows, the environment variables PATH and COMSPEC in any capitalization are converted automatically to upper case. For instance, the PATH variable could be exported by the operating system as ``path'', ``Path'', ``PaTh'', etc., causing otherwise simple Tcl code to have to support many special cases. All other environment variables inherited by Tcl are left unmodified. Setting an env array variable to blank is the same as unsetting it as this is the behavior of the underlying Windows OS. It should be noted that relying on an existing and empty environment variable will not work on Windows and is discouraged for cross-platform usage.

The following elements of env are special to Tcl:

env(HOME)

This environment variable, if set, gives the location of the directory considered to be the current user's home directory, and to which a call of cd without arguments or with just ``~'' as an argument will change into. Most platforms set this correctly by default; it does not normally need to be set by user code.

env(TCL_LIBRARY)

If set, then it specifies the location of the directory containing library scripts (the value of this variable will be assigned to the tcl_library variable and therefore returned by the command info library). If this variable is not set then a default value is used.

Note that this environment variable should not normally be set.

env(TCLLIBPATH)

If set, then it must contain a valid Tcl list giving directories to search during auto-load operations. Directories must be specified in Tcl format, using ``/'' as the path separator, regardless of platform. This variable is only used when initializing the auto_path variable.

env(TCL_TZ), env(TZ)

These specify the default timezone used for parsing and formatting times and dates in the clock command. On many platforms, the TZ environment variable is set up by the operating system.

env(LC_ALL), env(LC_MESSAGES), env(LANG)

These environment variables are used by the msgcat package to determine what locale to format messages using.

env(TCL_INTERP_DEBUG_FRAME)

If existing, it has the same effect as running interp debug {} -frame 1 as the very first command of each new Tcl interpreter.

errorCode

This variable holds the value of the -errorcode return option set by the most recent error that occurred in this interpreter. This list value represents additional information about the error in a form that is easy to process with programs. The first element of the list identifies a general class of errors, and determines the format of the rest of the list. The following formats for -errorcode return options are used by the Tcl core; individual applications may define additional formats.

ARITH code msg

This format is used when an arithmetic error occurs (e.g. an attempt to divide zero by zero in the expr command). Code identifies the precise error and msg provides a human-readable description of the error. Code will be either DIVZERO (for an attempt to divide by zero), DOMAIN (if an argument is outside the domain of a function, such as acos(-3)), IOVERFLOW (for integer overflow), OVERFLOW (for a floating-point overflow), or UNKNOWN (if the cause of the error cannot be determined).

Detection of these errors depends in part on the underlying hardware and system libraries.

CHILDKILLED pid sigName msg

This format is used when a child process has been killed because of a signal. The pid element will be the process's identifier (in decimal). The sigName element will be the symbolic name of the signal that caused the process to terminate; it will be one of the names from the include file signal.h, such as SIGPIPE. The msg element will be a short human-readable message describing the signal, such as ``write on pipe with no readers'' for SIGPIPE.

CHILDSTATUS pid code

This format is used when a child process has exited with a non-zero exit status. The pid element will be the process's identifier (in decimal) and the code element will be the exit code returned by the process (also in decimal).

CHILDSUSP pid sigName msg

This format is used when a child process has been suspended because of a signal. The pid element will be the process's identifier, in decimal. The sigName element will be the symbolic name of the signal that caused the process to suspend; this will be one of the names from the include file signal.h, such as SIGTTIN. The msg element will be a short human-readable message describing the signal, such as ``background tty read'' for SIGTTIN.

NONE

This format is used for errors where no additional information is available for an error besides the message returned with the error. In these cases the -errorcode return option will consist of a list containing a single element whose contents are NONE.

POSIX errName msg

If the first element is POSIX, then the error occurred during a POSIX kernel call. The errName element will contain the symbolic name of the error that occurred, such as ENOENT; this will be one of the values defined in the include file errno.h. The msg element will be a human-readable message corresponding to errName, such as ``no such file or directory'' for the ENOENT case.

TCL ...

Indicates some sort of problem generated in relation to Tcl itself, e.g. a failure to look up a channel or variable.

To set the -errorcode return option, applications should use library procedures such as Tcl_SetObjErrorCode, Tcl_SetReturnOptions, and Tcl_PosixError, or they may invoke the -errorcode option of the return command. If none of these methods for setting the error code has been used, the Tcl interpreter will reset the variable to NONE after the next error.

errorInfo

This variable holds the value of the -errorinfo return option set by the most recent error that occurred in this interpreter. This string value will contain one or more lines identifying the Tcl commands and procedures that were being executed when the most recent error occurred. Its contents take the form of a stack trace showing the various nested Tcl commands that had been invoked at the time of the error.

tcl_library

This variable holds the name of a directory containing the system library of Tcl scripts, such as those used for auto-loading. The value of this variable is returned by the info library command. See the library manual entry for details of the facilities provided by the Tcl script library. Normally each application or package will have its own application-specific script library in addition to the Tcl script library; each application should set a global variable with a name like $app_library (where app is the application's name) to hold the network file name for that application's library directory. The initial value of tcl_library is set when an interpreter is created by searching several different directories until one is found that contains an appropriate Tcl startup script. If the TCL_LIBRARY environment variable exists, then the directory it names is checked first. If TCL_LIBRARY is not set or doesn't refer to an appropriate directory, then Tcl checks several other directories based on a compiled-in default location, the location of the binary containing the application, and the current working directory.

tcl_patchLevel

When an interpreter is created Tcl initializes this variable to hold a string giving the current patch level for Tcl, such as 8.4.16 for Tcl 8.4 with the first sixteen official patches, or 8.5b3 for the third beta release of Tcl 8.5. The value of this variable is returned by the info patchlevel command.

tcl_pkgPath

This variable holds a list of directories indicating where packages are normally installed. It is not used on Windows. It typically contains either one or two entries; if it contains two entries, the first is normally a directory for platform-dependent packages (e.g., shared library binaries) and the second is normally a directory for platform-independent packages (e.g., script files). Typically a package is installed as a subdirectory of one of the entries in the tcl_pkgPath variable. The directories in the tcl_pkgPath variable are included by default in the auto_path variable, so they and their immediate subdirectories are automatically searched for packages during package require commands. Note: tcl_pkgPath is not intended to be modified by the application. Its value is added to auto_path at startup; changes to tcl_pkgPath are not reflected in auto_path. If you want Tcl to search additional directories for packages you should add the names of those directories to auto_path, not tcl_pkgPath.

tcl_platform

This is an associative array whose elements contain information about the platform on which the application is running, such as the name of the operating system, its current release number, and the machine's instruction set. The elements listed below will always be defined, but they may have empty strings as values if Tcl could not retrieve any relevant information. In addition, extensions and applications may add additional values to the array. The predefined elements are:

byteOrder

The native byte order of this machine: either littleEndian or bigEndian.

debug

If this variable exists, then the interpreter was compiled with and linked to a debug-enabled C run-time. This variable will only exist on Windows, so extension writers can specify which package to load depending on the C run-time library that is in use. This is not an indication that this core contains symbols.

engine

The name of the Tcl language implementation. When the interpreter is first created, this is always set to the string Tcl.

machine

The instruction set executed by this machine, such as intel, PPC, 68k, or sun4m. On UNIX machines, this is the value returned by uname -m.

os

The name of the operating system running on this machine, such as Windows NT or SunOS. On UNIX machines, this is the value returned by uname -s.

osVersion

The version number for the operating system running on this machine. On UNIX machines, this is the value returned by uname -r.

pathSeparator

The character that should be used to split PATH-like environment variables into their corresponding list of directory names.

platform

Either windows, or unix. This identifies the general operating environment of the machine.

pointerSize

This gives the size of the native-machine pointer in bytes (strictly, it is same as the result of evaluating sizeof(void*) in C.)

threaded

If this variable exists, then the interpreter was compiled with threads enabled.

user

This identifies the current user based on the login information available on the platform. This value comes from the getuid() and getpwuid() system calls on Unix, and the value from the GetUserName() system call on Windows.

wordSize

This gives the size of the native-machine word in bytes (strictly, it is same as the result of evaluating sizeof(long) in C.)

tcl_precision

This variable controls the number of digits to generate when converting floating-point values to strings. It defaults to 0. Applications should not change this value; it is provided for compatibility with legacy code.

The default value of 0 is special, meaning that Tcl should convert numbers using as few digits as possible while still distinguishing any floating point number from its nearest neighbours. It differs from using an arbitrarily high value for tcl_precision in that an inexact number like 1.4 will convert as 1.4 rather than 1.3999999999999999 even though the latter is nearer to the exact value of the binary number.

If tcl_precision is not zero, then when Tcl converts a floating point number, it creates a decimal representation of at most tcl_precision significant digits; the result may be shorter if the shorter result represents the original number exactly. If no result of at most tcl_precision digits is an exact representation of the original number, the one that is closest to the original number is chosen. If the original number lies precisely between two equally accurate decimal representations, then the one with an even value for the least significant digit is chosen; for instance, if tcl_precision is 3, then 0.3125 will convert to 0.312, not 0.313, while 0.6875 will convert to 0.688, not 0.687. Any string of trailing zeroes that remains is trimmed.

a tcl_precision value of 17 digits is ``perfect'' for IEEE floating-point in that it allows double-precision values to be converted to strings and back to binary with no loss of information. For this reason, you will often see it as a value in legacy code that must run on Tcl versions before 8.5. It is no longer recommended; as noted above, a zero value is the preferred method.

All interpreters in a thread share a single tcl_precision value: changing it in one interpreter will affect all other interpreters as well. Safe interpreters are not allowed to modify the variable.

Valid values for tcl_precision range from 0 to 17.

tcl_rcFileName

This variable is used during initialization to indicate the name of a user-specific startup file. If it is set by application-specific initialization, then the Tcl startup code will check for the existence of this file and source it if it exists. For example, for wish the variable is set to ~/.wishrc for Unix and ~/wishrc.tcl for Windows.

tcl_traceCompile

The value of this variable can be set to control how much tracing information is displayed during bytecode compilation. By default, tcl_traceCompile is zero and no information is displayed. Setting tcl_traceCompile to 1 generates a one-line summary in stdout whenever a procedure or top-level command is compiled. Setting it to 2 generates a detailed listing in stdout of the bytecode instructions emitted during every compilation. This variable is useful in tracking down suspected problems with the Tcl compiler.

This variable and functionality only exist if TCL_COMPILE_DEBUG was defined during Tcl's compilation.

tcl_traceExec

The value of this variable can be set to control how much tracing information is displayed during bytecode execution. By default, tcl_traceExec is zero and no information is displayed. Setting tcl_traceExec to 1 generates a one-line trace in stdout on each call to a Tcl procedure. Setting it to 2 generates a line of output whenever any Tcl command is invoked that contains the name of the command and its arguments. Setting it to 3 produces a detailed trace showing the result of executing each bytecode instruction. Note that when tcl_traceExec is 2 or 3, commands such as set and incr that have been entirely replaced by a sequence of bytecode instructions are not shown. Setting this variable is useful in tracking down suspected problems with the bytecode compiler and interpreter.

This variable and functionality only exist if TCL_COMPILE_DEBUG was defined during Tcl's compilation.

tcl_wordchars

The value of this variable is a regular expression that can be set to control what are considered ``word'' characters, for instances like selecting a word by double-clicking in text in Tk. It is platform dependent. On Windows, it defaults to \S, meaning anything but a Unicode space character. Otherwise it defaults to \w, which is any Unicode word character (number, letter, or underscore).

tcl_nonwordchars

The value of this variable is a regular expression that can be set to control what are considered ``non-word'' characters, for instances like selecting a word by double-clicking in text in Tk. It is platform dependent. On Windows, it defaults to \s, meaning any Unicode space character. Otherwise it defaults to \W, which is anything but a Unicode word character (number, letter, or underscore).

tcl_version

When an interpreter is created Tcl initializes this variable to hold the version number for this version of Tcl in the form x.y. Changes to x represent major changes with probable incompatibilities and changes to y represent small enhancements and bug fixes that retain backward compatibility. The value of this variable is returned by the info tclversion command.

OTHER GLOBAL VARIABLES

The following variables are only guaranteed to exist in tclsh and wish executables; the Tcl library does not define them itself but many Tcl environments do.

argc

The number of arguments to tclsh or wish.

argv

Tcl list of arguments to tclsh or wish.

argv0

The script that tclsh or wish started executing (if it was specified) or otherwise the name by which tclsh or wish was invoked.

tcl_interactive

Contains 1 if tclsh or wish is running interactively (no script was specified and standard input is a terminal-like device), 0 otherwise.

EXAMPLES

To add a directory to the collection of locations searched by package require, e.g., because of some application-specific packages that are used, the auto_path variable needs to be updated:

lappend ::auto_path [file join [pwd] "theLibDir"]

A simple though not very robust way to handle command line arguments of the form ``-foo 1 -bar 2'' is to load them into an array having first loaded in the default settings:

array set arguments {-foo 0 -bar 0 -grill 0}
array set arguments $::argv
puts "foo is $arguments(-foo)"
puts "bar is $arguments(-bar)"
puts "grill is $arguments(-grill)"

The argv0 global variable can be used (in conjunction with the info script command) to determine whether the current script is being executed as the main script or loaded as a library. This is useful because it allows a single script to be used as both a library and a demonstration of that library:

if {$::argv0 eq [info script]} {
    # running as: tclsh example.tcl
} else {
    package provide Example 1.0
}

KEYWORDS

arithmetic, bytecode, compiler, error, environment, POSIX, precision, subprocess, user, variables