Methods in class Tcl¶

To create a new Tcl interpreter, use

    $interp = Tcl->new;

The following methods and routines can then be used on the Perl object returned (the object argument omitted in each case).

$interp->Init ()

Invoke Tcl_Init on the interpreter.

$interp->CreateSlave (NAME, SAFE)

Invoke Tcl_CreateSlave on the interpreter. Name is arbitrary. The safe variable, if true, creates a safe sandbox interpreter.
See: http://www.tcl.tk/software/plugin/safetcl.html
http://www.tcl.tk/man/tcl8.4/TclCmd/safe.htm

This command returns a new interpreter.

$interp->Eval (STRING, FLAGS)

Evaluate script STRING in the interpreter. If the script returns successfully (TCL_OK) then the Perl return value corresponds to Tcl interpreter's result otherwise a die exception is raised with the $@ variable corresponding to Tcl's interpreter result object. In each case, corresponds means that if the method is called in scalar context then the string result is returned but if the method is called in list context then the result is split as a Tcl list and returned as a Perl list. The FLAGS field is optional and can be a bitwise OR of the constants Tcl::EVAL_GLOBAL or Tcl::EVAL_DIRECT.

$interp->GlobalEval (STRING)

REMOVED. Evalulate script STRING at global level. Call Eval(STRING, Tcl::EVAL_GLOBAL) instead.

$interp->EvalFile (FILENAME)

Evaluate the contents of the file with name FILENAME. Otherwise, the same as Eval() above.

$interp->EvalFileHandle (FILEHANDLE)

Evaluate the contents of the Perl filehandle FILEHANDLE. Otherwise, the same as Eval() above. Useful when using the filehandle DATA to tack on a Tcl script following an __END__ token.

$interp->call (PROC, ARG, ...)

Looks up procedure PROC in the interpreter and invokes it using Tcl's eval semantics that does command tracing and will use the ::unknown (AUTOLOAD) mechanism. The arguments (ARG, ...) are not passed through the Tcl parser. For example, spaces embedded in any ARG will not cause it to be split into two Tcl arguments before being passed to PROC.

Before invoking procedure PROC special processing is performed on ARG list:

1. All subroutine references within ARG will be substituted with Tcl name which is responsible to invoke this subroutine. This Tcl name will be created using CreateCommand subroutine (see below).

2. All references to scalars will be substituted with names of Tcl variables transformed appropriately.

These first two items allow one to write and expect it to work properly such code as:

  my $r = 'aaaa';
  button(".d", -textvariable => \$r, -command=>sub {$r++});
    

3. All references to hashes will be substituted with names of Tcl array variables transformed appropriately.

4. As a special case, there is a mechanism to deal with Tk's special event variables (they are mentioned as '%x', '%y' and so on throughout Tcl). When creating a subroutine reference that uses such variables, you must declare the desired variables using Tcl::Ev as the first argument to the subroutine. Example:

  sub textPaste {
      my ($x,$y,$w) = @_;
      widget($w)->insert("\@$x,$y", $interp->Eval('selection get'));
  }
  $widget->bind('<2>', [\&textPaste, Tcl::Ev('%x', '%y'), $widget] );
    

$interp->return_ref (NAME)

returns a reference corresponding to NAME, which was associated during previously called "$interpnt->call(...)" preprocessing. As a typical example this could be variable associated with a widget.

$interp->delete_ref (NAME)

deletes and returns a reference corresponding to NAME, which was associated during previously called "$interpnt->call(...)" preprocessing. this follows _code_dispose about if NAME is a TClNAME or a DESCNAME. if NAME is a VARNAME then it is just deleted and returned.

$interp->icall (PROC, ARG, ...)

Looks up procedure PROC in the interpreter and invokes it using Tcl's eval semantics that does command tracing and will use the ::unknown (AUTOLOAD) mechanism. The arguments (ARG, ...) are not passed through the Tcl parser. For example, spaces embedded in any ARG will not cause it to be split into two Tcl arguments before being passed to PROC.

This is the lower-level procedure that the 'call' method uses. Arguments are converted efficiently from Perl SVs to Tcl_Objs. A Perl AV array becomes a Tcl_ListObj, an SvIV becomes a Tcl_IntObj, etc. The reverse conversion is done to the result.

$interp->invoke (PROC, ARG, ...)

Looks up procedure PROC in the interpreter and invokes it directly with arguments (ARG, ...) without passing through the Tcl parser. For example, spaces embedded in any ARG will not cause it to be split into two Tcl arguments before being passed to PROC. This differs from icall/call in that it directly invokes the command name without allowing for command tracing or making use of Tcl's unknown (AUTOLOAD) mechanism. If the command does not already exist in the interpreter, and error will be thrown.

Arguments are converted efficiently from Perl SVs to Tcl_Objs. A Perl AV array becomes a Tcl_ListObj, an SvIV becomes a Tcl_IntObj, etc. The reverse conversion is done to the result.

Tcl::Ev (FIELD, ...)

Used to declare %-substitution variables of interest to a subroutine callback. FIELD is expected to be of the form "%#" where # is a single character, and multiple fields may be specified. Returns a blessed object that the 'call' method will recognize when it is passed as the first argument to a subroutine in a callback. See description of 'call' method for details.

$interp->result ()

Returns the current Tcl interpreter result. List v. scalar context is handled as in Eval() above.

$interp->CreateCommand (CMDNAME, CMDPROC, CLIENTDATA, DELETEPROC, FLAGS)

Binds a new procedure named CMDNAME into the interpreter. The CLIENTDATA and DELETEPROC arguments are optional. There are two cases:

(1) CMDPROC is the address of a C function

(presumably obtained using dl_open and dl_find_symbol. In this case CLIENTDATA and DELETEPROC are taken to be raw data of the ClientData and deleteProc field presumably obtained in a similar way.

(2) CMDPROC is a Perl subroutine

(either a sub name, a sub reference or an anonymous sub). In this case CLIENTDATA can be any perl scalar (e.g. a ref to some other data) and DELETEPROC must be a perl sub too. When CMDNAME is invoked in the Tcl interpreter, the arguments passed to the Perl sub CMDPROC are

    (CLIENTDATA, INTERP, LIST)
    

where INTERP is a Perl object for the Tcl interpreter which called out and LIST is a Perl list of the arguments CMDNAME was called with. If the 1-bit of FLAGS is set then the 3 first arguments on the call to CMDPROC are suppressed. As usual in Tcl, the first element of the list is CMDNAME itself. When CMDNAME is deleted from the interpreter (either explicitly with DeleteCommand or because the destructor for the interpreter object is called), it is passed the single argument CLIENTDATA.

$interp->DeleteCommand (CMDNAME)

Deletes command CMDNAME from the interpreter. If the command was created with a DELETEPROC (see CreateCommand above), then it is invoked at this point. When a Tcl interpreter object is destroyed either explicitly or implicitly, an implicit DeleteCommand happens on all its currently registered commands.

(TCLNAME,GENCODE) = $interp->create_tcl_sub(CODEREF, EVENTS, TCLNAME, DESCRNAME)

Creates a COMMAND called TCLNAME calling CODEREF in the interpreter $interp, and adds it to the internal tracking structure. DESCRNAME and TCLNAME should not begin with = or @ or ::perl:: to avoid colisions If TCLNAME IS blank or undef, it is constructed from the CODEREF address. GENCODE starts as TCLNAME but gets @$EVENTS which can contain %vars joined to it.

TCLNAME and DESCRNAME get stored in an internal structure, and can be used to purge things fRom the command table via code_destroy or $interp->delete_ref;

Returns (TCLNAME,GENCODE). if you are creating code refs with this you can continue to use the same coderef and it will be converted on each call. but if you save GENCODE, you can replace the anon-coderef call in the tcl command with GENCODE.

for instance

  $interp->call('FILEEVENT',$fileref,WRITABLE=>sub {...});
    

can be replaced by

  my ($tclcode,$gencode)=$interp->create_tcl_sub(sub{...}, EVENTS, TCLNAME, DESCRNAME);
  $interp->call('FILEEVENT',$gencode,WRITABLE=>$gencode);
    

or

  my $sub=sub{....};
  $interp->call('FILEEVENT',$fileref,WRITABLE=>$sub);
    

can be replaced by

  my ($tclcode,$gencode)=$interp->create_tcl_sub($sub, EVENTS, TCLNAME, DESCRNAME);
  $interp->call('FILEEVENT',$gencode,WRITABLE=>$gencode);
    

although

  $interp->call('FILEEVENT',$fileref,WRITABLE=>$sub);
    

will stil work fine too.

Then you later call

  $interp->delete_ref($tclname);
    

when you are finished with that sub to clean it from the internal tracking and command table. This means no automatic cleanup will occur on the sub{...} or $sub

And after the destroy inside Tcl any triggering writable on $fileref will fail as well. so it should be replaced first via
$interp->call('FILEEVENT',$fileref,WRITABLE=>'');

(CODEREF) = Tcl::_code_dispose(NAME)

Purges the internal table of a NAME and may initiate destruction of something created thru call or create_tcl_sub

TCLNAME and DESCRNAME get stored in an internal structure, and can be used to purge things form the command table. calling _code_dispose on a TCLNAME retruned from create_tcl_sub removes all use instances and purges the command table. calling _code_dispose on a DESCRNAME passed to create_tcl_sub removes only that instace Code used in a DESCRNAME may be used in other places as well, only when the last usage is purged does the entry get purged from the command table

While the internal tracking structure saves the INTERP the code was added to, it itself does not keep things separated by INTERP, A TCLNAME or DESCRNAMe can only exist in one INTERP at a time, using a new INTERP just causes the one in the last INTERP to disappear, and probably end up with the Tcl code getting deleted

Returns (CODEREF), this is the original coderef

$interp->SetResult (STRING)

Sets Tcl interpreter result to STRING.

$interp->AppendResult (LIST)

Appends each element of LIST to Tcl's interpreter result object.

$interp->AppendElement (STRING)

Appends STRING to Tcl interpreter result object as an extra Tcl list element.

$interp->ResetResult ()

Resets Tcl interpreter result.

$interp->SplitList (STRING)

Splits STRING as a Tcl list. Returns a Perl list or the empty list if there was an error (i.e. STRING was not a properly formed Tcl list). In the latter case, the error message is left in Tcl's interpreter result object.

$interp->SetVar (VARNAME, VALUE, FLAGS)

The FLAGS field is optional. Sets Tcl variable VARNAME in the interpreter to VALUE. The FLAGS argument is the usual Tcl one and can be a bitwise OR of the constants Tcl::GLOBAL_ONLY, Tcl::LEAVE_ERR_MSG, Tcl::APPEND_VALUE, Tcl::LIST_ELEMENT.

$interp->SetVar2 (VARNAME1, VARNAME2, VALUE, FLAGS)

Sets the element VARNAME1(VARNAME2) of a Tcl array to VALUE. The optional argument FLAGS behaves as in SetVar above.

$interp->GetVar (VARNAME, FLAGS)

Returns the value of Tcl variable VARNAME. The optional argument FLAGS behaves as in SetVar above.

$interp->GetVar2 (VARNAME1, VARNAME2, FLAGS)

Returns the value of the element VARNAME1(VARNAME2) of a Tcl array. The optional argument FLAGS behaves as in SetVar above.

$interp->UnsetVar (VARNAME, FLAGS)

Unsets Tcl variable VARNAME. The optional argument FLAGS behaves as in SetVar above.

$interp->UnsetVar2 (VARNAME1, VARNAME2, FLAGS)

Unsets the element VARNAME1(VARNAME2) of a Tcl array. The optional argument FLAGS behaves as in SetVar above.

Command table cleanup¶

In V1.03 command table cleanup was intoduced. This tries to keep the internal structure and command table clean. In V1.02 and prior heavy use of sub { .. } in Tcl commands could pollute these tables as they were never cleared. Command table cleanup tries to alieviate this.

if you call create_tcl_sub the internal reference exists until you delete_ref or _code_dispose it, or you call create_tcl_sub with the same DESCRNAME.

if the internal reference was created internaly by call(...) there are two rules

1)

If the command is an "after" the internal references is keept at least until 1 second after the delay. If there are still other "users" of the TCLNAME then it is not deleted until the last one goes away. If another call with the same CODEREF happens before this, then it will get registered as a "user" without any need to delete/recreate the tcl command first.

2)

otherwise a DESCRNAME is created with the text sections of the command, prefaced by "=". Like "=after 1000" or "=:.m.m add command -command -label Exit" or "=::button .f3.b8 -text conn -command" or "=gets sock9ac2b50" or "=fileevent sock9827430 writable"

the TCLCODES created for that command will be kept at least until a command with the same DESCRNAME and containing a subroutine reference is run again. Since many DESCRNAMES can reference the same TCLNAME only when the last DESCRNAME referencing a TCLNAME is released is the TCLNAME purged.

NOTE: Since
$interp->call('fileevent','sock9827430','writable'); does not contain a subroutine reference, it will not release/free the TCLNAME/DESCRNAME created by
$interp->call('fileevent','sock9827430','writable',sub{...}); even though that is the way you deactivate a writable/readable callback in Tcl.

Prior to V1.06 there was also a problem with the coderef never getting cleared from sas, a refcount was kept at the PVCV that prevented it from getting garbage collected, but that SV itself got "lost" and could never be garbage collected, thereby also keeping anything in that codes PAD.

To assist in tracking chages to the internal table and the commands table 3 trace subs were added, set them to non-blank or non-zero to add the tracking output to SYSOUT, like this in your code:

    sub Tcl::TRACE_SHOWCODE(){1}

Tcl::TRACE_SHOWCODE

Display all generated Tcl code by call(). Be aware: Tkx::MainLoop runs by issuing a lot of "winfo exists ." calls, a LOT. But this is a nice way to tell what your programs are doing to Tcl.

Tcl::TRACE_CREATECOMMAND

Display Tcl subroutine creation by call/create_tcl_sub

Tcl::TRACE_DELETECOMMAND

Display Tcl subroutine deletion by cleanup/delete_ref/_code_dispose

Linking Perl and Tcl variables¶

You can tie a Perl variable (scalar or hash) into class Tcl::Var so that changes to a Tcl variable automatically "change" the value of the Perl variable. In fact, as usual with Perl tied variables, its current value is just fetched from the Tcl variable when needed and setting the Perl variable triggers the setting of the Tcl variable.

To tie a Perl scalar $scalar to the Tcl variable tclscalar in interpreter $interp with optional flags $flags (see SetVar above), use

        tie $scalar, "Tcl::Var", $interp, "tclscalar", $flags;

Omit the $flags argument if not wanted.

To tie a Perl hash %hash to the Tcl array variable array in interpreter $interp with optional flags $flags (see SetVar above), use

        tie %hash, "Tcl::Var", $interp, "array", $flags;

Omit the $flags argument if not wanted. Any alteration to Perl variable $hash{"key"} affects the Tcl variable array(key) and vice versa.

Accessing Perl from within Tcl¶

After creation of Tcl interpreter, in addition to evaluation of Tcl/Tk commands within Perl, other way round also instantiated. Within a special namespace " ::perl " following objects are created:

   ::perl::Eval

So it is possible to use Perl objects from within Tcl.

Moving Tcl/Tk around with Tcl.pm¶

NOTE: explanations below is for developers managing Tcl/Tk installations itself, users should skip this section.

In order to create Tcl/Tk application with this module, you need to make sure that Tcl/Tk is available within visibility of this module. There are many ways to achieve this, varying on ease of starting things up and providing flexible moveable archived files.

Following list enumerates them, in order of increased possibility to change location.

• First method

  Install Tcl/Tk first, then install Perl module Tcl, so installed Tcl/Tk will be used. This is most normal approach, and no care of Tcl/Tk distribution is taken on Perl side (this is done on Tcl/Tk side)

• Second method

  Copy installed Tcl/Tk binaries to some location, then install Perl module Tcl with a special action to make Tcl.pm know of this location. This approach makes sure that only chosen Tcl installation is used.

• Third method

  During compiling Tcl Perl module, Tcl/Tk could be statically linked into module's shared library and all other files zipped into a single archive, so each file extracted when needed.

  To link Tcl/Tk binaries, prepare their libraries and then instruct Makefile.PL to use these libraries in a link stage. (TODO provide better detailed description)