Documentation generated from fossil trunk
namespace -
create and manipulate contexts for commands and variables
namespace ?subcommand? ?arg ...?
The namespace command lets you create, access, and destroy separate contexts for commands and variables. See the section WHAT IS A NAMESPACE? below for a brief overview of namespaces. The legal values of subcommand are listed below. Note that you can abbreviate the subcommands.
If namespace has leading namespace qualifiers and any leading namespaces do not exist, they are automatically created.
When pattern arguments are present, each pattern is a qualified name like foo::x or a::p*. That is, it includes the name of an exporting namespace and may have glob-style special characters in the command name at the end of the qualified name. Glob characters may not appear in a namespace name. When the namespace name is not fully qualified (i.e., does not start with a namespace separator) it is resolved as a namespace name in the way described in the NAME RESOLUTION section; it is an error if no namespace with that name can be found.
All the commands that match a pattern string and which are currently exported from their namespace are added to the current namespace. This is done by creating a new command in the current namespace that points to the exported command in its original namespace; when the new imported command is called, it invokes the exported command. This command normally returns an error if an imported command conflicts with an existing command. However, if the -force option is given, imported commands will silently replace existing commands. The namespace import command has snapshot semantics: that is, only requested commands that are currently defined in the exporting namespace are imported. In other words, you can import only the commands that are in a namespace at the time when the namespace import command is executed. If another command is defined and exported in this namespace later on, it will not be imported.
namespace inscope ::foo $script $x $y $z
is equivalent to
namespace eval ::foo [concat $script [list $x $y $z]]
thus additional arguments will not undergo a second round of substitution, as is the case with namespace eval.
A namespace is a collection of commands and variables. It encapsulates the commands and variables to ensure that they will not interfere with the commands and variables of other namespaces. Tcl has always had one such collection, which we refer to as the global namespace. The global namespace holds all global variables and commands. The namespace eval command lets you create new namespaces. For example,
namespace eval Counter {
namespace export bump
variable num 0
proc bump {} {
variable num
incr num
}
}
creates a new namespace containing the variable num and the procedure bump. The commands and variables in this namespace are separate from other commands and variables in the same program. If there is a command named bump in the global namespace, for example, it will be different from the command bump in the Counter namespace.
Namespace variables resemble global variables in Tcl. They exist outside of the procedures in a namespace but can be accessed in a procedure via the variable command, as shown in the example above.
Namespaces are dynamic. You can add and delete commands and variables at any time, so you can build up the contents of a namespace over time using a series of namespace eval commands. For example, the following series of commands has the same effect as the namespace definition shown above:
namespace eval Counter {
variable num 0
proc bump {} {
variable num
return [incr num]
}
}
namespace eval Counter {
proc test {args} {
return $args
}
}
namespace eval Counter {
rename test ""
}
Note that the test procedure is added to the Counter namespace, and later removed via the rename command.
Namespaces can have other namespaces within them, so they nest hierarchically. A nested namespace is encapsulated inside its parent namespace and can not interfere with other namespaces.
Each namespace has a textual name such as history or ::safe::interp. Since namespaces may nest, qualified names are used to refer to commands, variables, and child namespaces contained inside namespaces. Qualified names are similar to the hierarchical path names for Unix files or Tk widgets, except that :: is used as the separator instead of / or .. The topmost or global namespace has the name "" (i.e., an empty string), although :: is a synonym. As an example, the name ::safe::interp::create refers to the command create in the namespace interp that is a child of namespace ::safe, which in turn is a child of the global namespace, ::.
If you want to access commands and variables from another namespace, you must use some extra syntax. Names must be qualified by the namespace that contains them. From the global namespace, we might access the Counter procedures like this:
Counter::bump 5 Counter::Reset
We could access the current count like this:
puts "count = $Counter::num"
When one namespace contains another, you may need more than one qualifier to reach its elements. If we had a namespace Foo that contained the namespace Counter, you could invoke its bump procedure from the global namespace like this:
Foo::Counter::bump 3
You can also use qualified names when you create and rename commands. For example, you could add a procedure to the Foo namespace like this:
proc Foo::Test {args} {return $args}
And you could move the same procedure to another namespace like this:
rename Foo::Test Bar::Test
There are a few remaining points about qualified names that we should cover. Namespaces have nonempty names except for the global namespace. :: is disallowed in simple command, variable, and namespace names except as a namespace separator. Extra colons in any separator part of a qualified name are ignored; i.e. two or more colons are treated as a namespace separator. A trailing :: in a qualified variable or command name refers to the variable or command named {}. However, a trailing :: in a qualified namespace name is ignored.
In general, all Tcl commands that take variable and command names support qualified names. This means you can give qualified names to such commands as set, proc, rename, and interp alias. If you provide a fully-qualified name that starts with a ::, there is no question about what command, variable, or namespace you mean. However, if the name does not start with a :: (i.e., is relative), Tcl follows basic rules for looking it up:
In the following example,
set traceLevel 0
namespace eval Debug {
printTrace $traceLevel
}
Tcl looks for traceLevel in the namespace Debug and then in the global namespace. It looks up the command printTrace in the same way. If a variable or command name is not found in either context, the name is undefined. To make this point absolutely clear, consider the following example:
set traceLevel 0
namespace eval Foo {
variable traceLevel 3
namespace eval Debug {
printTrace $traceLevel
}
}
Here Tcl looks for traceLevel first in the namespace Foo::Debug. Since it is not found there, Tcl then looks for it in the global namespace. The variable Foo::traceLevel is completely ignored during the name resolution process.
You can use the namespace which command to clear up any question about name resolution. For example, the command:
namespace eval Foo::Debug {namespace which -variable traceLevel}
returns ::traceLevel. On the other hand, the command,
namespace eval Foo {namespace which -variable traceLevel}
returns ::Foo::traceLevel.
As mentioned above, namespace names are looked up differently than the names of variables and commands. Namespace names are always resolved in the current namespace. This means, for example, that a namespace eval command that creates a new namespace always creates a child of the current namespace unless the new namespace name begins with ::.
Tcl has no access control to limit what variables, commands, or namespaces you can reference. If you provide a qualified name that resolves to an element by the name resolution rule above, you can access the element.
You can access a namespace variable from a procedure in the same namespace by using the variable command. Much like the global command, this creates a local link to the namespace variable. If necessary, it also creates the variable in the current namespace and initializes it. Note that the global command only creates links to variables in the global namespace. It is not necessary to use a variable command if you always refer to the namespace variable using an appropriate qualified name.
Namespaces are often used to represent libraries. Some library commands are used so frequently that it is a nuisance to type their qualified names. For example, suppose that all of the commands in a package like BLT are contained in a namespace called Blt. Then you might access these commands like this:
Blt::graph .g -background red Blt::table . .g 0,0
If you use the graph and table commands frequently, you may want to access them without the Blt:: prefix. You can do this by importing the commands into the current namespace, like this:
namespace import Blt::*
This adds all exported commands from the Blt namespace into the current namespace context, so you can write code like this:
graph .g -background red table . .g 0,0
The namespace import command only imports commands from a namespace that that namespace exported with a namespace export command.
Importing every command from a namespace is generally a bad idea since you do not know what you will get. It is better to import just the specific commands you need. For example, the command
namespace import Blt::graph Blt::table
imports only the graph and table commands into the current context.
If you try to import a command that already exists, you will get an error. This prevents you from importing the same command from two different packages. But from time to time (perhaps when debugging), you may want to get around this restriction. You may want to reissue the namespace import command to pick up new commands that have appeared in a namespace. In that case, you can use the -force option, and existing commands will be silently overwritten:
namespace import -force Blt::graph Blt::table
If for some reason, you want to stop using the imported commands, you can remove them with a namespace forget command, like this:
namespace forget Blt::*
This searches the current namespace for any commands imported from Blt. If it finds any, it removes them. Otherwise, it does nothing. After this, the Blt commands must be accessed with the Blt:: prefix.
When you delete a command from the exporting namespace like this:
rename Blt::graph ""
the command is automatically removed from all namespaces that import it.
You can export commands from a namespace like this:
namespace eval Counter {
namespace export bump reset
variable Num 0
variable Max 100
proc bump {{by 1}} {
variable Num
incr Num $by
Check
return $Num
}
proc reset {} {
variable Num
set Num 0
}
proc Check {} {
variable Num
variable Max
if {$Num > $Max} {
error "too high!"
}
}
}
The procedures bump and reset are exported, so they are included when you import from the Counter namespace, like this:
namespace import Counter::*
However, the Check procedure is not exported, so it is ignored by the import operation.
The namespace import command only imports commands that were declared as exported by their namespace. The namespace export command specifies what commands may be imported by other namespaces. If a namespace import command specifies a command that is not exported, the command is not imported.
The namespace code command is the means by which a script may be packaged for evaluation in a namespace other than the one in which it was created. It is used most often to create event handlers, Tk bindings, and traces for evaluation in the global context. For instance, the following code indicates how to direct a variable trace callback into the current namespace:
namespace eval a {
variable b
proc theTraceCallback { n1 n2 op } {
upvar 1 $n1 var
puts "the value of $n1 has changed to $var"
return
}
trace add variable b write [namespace code theTraceCallback]
}
set a::b c
When executed, it prints the message:
the value of a::b has changed to c
The namespace ensemble is used to create and manipulate ensemble commands, which are commands formed by grouping subcommands together. The commands typically come from the current namespace when the ensemble was created, though this is configurable. Note that there may be any number of ensembles associated with any namespace (including none, which is true of all namespaces by default), though all the ensembles associated with a namespace are deleted when that namespace is deleted. The link between an ensemble command and its namespace is maintained however the ensemble is renamed.
Three subcommands of the namespace ensemble command are defined:
When called, an ensemble command takes its first argument and looks it up (according to the rules described below) to discover a list of words to replace the ensemble command and subcommand with. The resulting list of words is then evaluated (with no further substitutions) as if that was what was typed originally (i.e. by passing the list of words through Tcl_EvalObjv) and returning the result of the command. Note that it is legal to make the target of an ensemble rewrite be another (or even the same) ensemble command. The ensemble command will not be visible through the use of the uplevel or info level commands.
The following options, supported by the namespace ensemble create and namespace ensemble configure commands, control how an ensemble command behaves:
The following extra option is allowed by namespace ensemble create:
The following extra option is allowed by namespace ensemble configure:
If an unknown handler is specified for an ensemble, that handler is called when the ensemble command would otherwise return an error due to it being unable to decide which subcommand to invoke. The exact conditions under which that occurs are controlled by the -subcommands, -map and -prefixes options as described above.
To execute the unknown handler, the ensemble mechanism takes the specified -unknown option and appends each argument of the attempted ensemble command invocation (including the ensemble command itself, expressed as a fully qualified name). It invokes the resulting command in the scope of the attempted call. If the execution of the unknown handler terminates normally, the ensemble engine reparses the subcommand (as described below) and tries to dispatch it again, which is ideal for when the ensemble's configuration has been updated by the unknown subcommand handler. Any other kind of termination of the unknown handler is treated as an error.
The result of the unknown handler is expected to be a list (it is an error if it is not). If the list is an empty list, the ensemble command attempts to look up the original subcommand again and, if it is not found this time, an error will be generated just as if the -unknown handler was not there (i.e. for any particular invocation of an ensemble, its unknown handler will be called at most once.) This makes it easy for the unknown handler to update the ensemble or its backing namespace so as to provide an implementation of the desired subcommand and reparse.
When the result is a non-empty list, the words of that list are used to replace the ensemble command and subcommand, just as if they had been looked up in the -map. It is up to the unknown handler to supply all namespace qualifiers if the implementing subcommand is not in the namespace of the caller of the ensemble command. Also note that when ensemble commands are chained (e.g. if you make one of the commands that implement an ensemble subcommand into an ensemble, in a manner similar to the text widget's tag and mark subcommands) then the rewrite happens in the context of the caller of the outermost ensemble. That is to say that ensembles do not in themselves place any namespace contexts on the Tcl call stack.
Where an empty -unknown handler is given (the default), the ensemble command will generate an error message based on the list of commands that the ensemble has defined (formatted similarly to the error message from Tcl_GetIndexFromObj). This is the error that will be thrown when the subcommand is still not recognized during reparsing. It is also an error for an -unknown handler to delete its namespace.
Create a namespace containing a variable and an exported command:
namespace eval foo {
variable bar 0
proc grill {} {
variable bar
puts "called [incr bar] times"
}
namespace export grill
}
Call the command defined in the previous example in various ways.
# Direct call
::foo::grill
# Use the command resolution path to find the name
namespace eval boo {
namespace path ::foo
grill
}
# Import into current namespace, then call local alias
namespace import foo::grill
grill
# Create two ensembles, one with the default name and one with a
# specified name. Then call through the ensembles.
namespace eval foo {
namespace ensemble create
namespace ensemble create -command ::foobar
}
foo grill
foobar grill
Look up where the command imported in the previous example came from:
puts "grill came from [namespace origin grill]"
Remove all imported commands from the current namespace:
namespace forget {*}[namespace import]
Create an ensemble for simple working with numbers, using the -parameters option to allow the operator to be put between the first and second arguments.
namespace eval do {
namespace export *
namespace ensemble create -parameters x
proc plus {x y} {expr { $x + $y }}
proc minus {x y} {expr { $x - $y }}
}
# In use, the ensemble works like this:
puts [do 1 plus [do 9 minus 7]]
interp(n), upvar(n), variable(n)