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.
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
Setting an element of the array will modify the corresponding
environment variable or create a new one if it doesn't already
Unsetting an element of env will remove the corresponding
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
- 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 won't work on windows and is discouraged for cross-platform usage.
After an error has occurred, this variable will be set to hold
a list value representing 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 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 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).
- CHILDKILLED pid sigName msg
- This format is used when a child process has been killed because of a signal. The second element of errorCode will be the process's identifier (in decimal). The third 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 fourth 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 second element of errorCode will be the process's identifier (in decimal) and the third 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 second element of errorCode will be the process's identifier, in decimal. The third 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 fourth element will be a short human-readable message describing the signal, such as ``background tty read'' for SIGTTIN.
- 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 errorCode will consist of a list containing a single element whose contents are NONE.
- POSIX errName msg
- If the first element of errorCode is POSIX, then the error occurred during a POSIX kernel call. The second element of the list 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 third element of the list will be a human-readable message corresponding to errName, such as ``no such file or directory'' for the ENOENT case.
To set errorCode, applications should use library procedures such as Tcl_SetErrorCode and Tcl_PosixError, or they may invoke the error command. If one of these methods hasn't been used, then the Tcl interpreter will reset the variable to NONE after the next error.
- After an error has occurred, this string 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.
- 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 isn't 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.
When an interpreter is created Tcl initializes this variable to
hold a string giving the current patch level for Tcl, such as
7.3p2 for Tcl 7.3 with the first two official patches, or
7.4b4 for the fourth beta release of Tcl 7.4.
The value of this variable is returned by the info patchlevel
- 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 $tcl_pkgPath. The directories in $tcl_pkgPath 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 it 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.
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 couldn't
retrieve any relevant information. In addition, extensions
and applications may add additional values to the array. The
predefined elements are:
- The native byte order of this machine: either littleEndian or bigEndian.
- 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.
- 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.
- The name of the operating system running on this machine, such as Windows 95, Windows NT, MacOS, or SunOS. On UNIX machines, this is the value returned by uname -s. On Windows 95 and Windows 98, the value returned will be Windows 95 to provide better backwards compatibility to Windows 95; to distinguish between the two, check the osVersion.
- The version number for the operating system running on this machine. On UNIX machines, this is the value returned by uname -r. On Windows 95, the version will be 4.0; on Windows 98, the version will be 4.10.
- Either windows or unix. This identifies the general operating environment of the machine.
- If this variable exists, then the interpreter was compiled with threads enabled.
- This identifies the current user based on the login information available on the platform. This comes from the USER or LOGNAME environment variable on Unix, and the value from GetUserName on Windows.
- This gives the size of the native-machine word in bytes (strictly, it is same as the result of evaluating sizeof(long) in C.)
This variable controls the number of digits to generate
when converting floating-point values to strings. It defaults
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. However, using 17 digits prevents
any rounding, which produces longer, less intuitive results. For example,
expr 1.4 returns 1.3999999999999999 with tcl_precision
set to 17, vs. 1.4 if tcl_precision is 12.
- All interpreters in a process share a single tcl_precision value: changing it in one interpreter will affect all other interpreters as well. However, safe interpreters are not allowed to modify the variable.
- 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.
- 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. It is also occasionally useful when converting existing code to use Tcl8.0.
This variable and functionality only exist if TCL_COMPILE_DEBUG was defined during Tcl's compilation.
- 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. It is also occasionally useful when converting code to use Tcl8.0.
This variable and functionality only exist if TCL_COMPILE_DEBUG was defined during Tcl's compilation.
- 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).
- 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).
- 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 VARIABLESThe 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.
- The number of arguments to tclsh or wish.
- Tcl list of arguments to tclsh or wish.
- The script that tclsh or wish started executing (if it was specified) or otherwise the name by which tclsh or wish was invoked.
- Contains 1 if tclsh or wish is running interactively (no script was specified and standard input is a terminal-like device), 0 otherwise.
The wish executably additionally specifies the following global variable:
If set, contains the user-supplied geometry specification to use for
the main Tk window.
KEYWORDSarithmetic, bytecode, compiler, error, environment, POSIX, precision, subprocess, variables