User Tools

Site Tools


en:projects:csp4th:readme

cspForth for Linux

Copyright (C) 2008 Manfred Mahlow (manfred.mahlow@forth-ev.de)

A first development snapshot of cspForth was released for the annual general meeting of the German Forth-Gesellschaft e.V. at Kloster Roggenburg, in April 2008. Now the second snapshot csp4th-sn02-080816.tar.gz is available and should be used. The documentation was updated accordingly.

Introduction

cspForth is a 32 bit Forth System for Linux on x86 PC-Systems. It came into being as a test bed, to evaluate the benefits of using OOP concepts with Forth. When looking for a small, easy to understand and easy to modify Forth System, I found that Reva Forth by Ron Aron was close to my needs, even though it did not comform to the ANS Forth Standard.

cspForth supports on demand loading of source code modules, importing functions from shared libraries and creating standalone applications. All this was inherited from Reva 6.0 and modified with respect to my needs.

cspForth has two special features, that make it different from most other Forth-Systems:

  • cspForth supports two search orders, the well known search order for vocabularies and wordlists and an extra search order for classes, objects and interfaces.
  • cspForth supports preludes. A prelude is a Forth word, that is assigned to another Forth word in a way, that it is executed before the word, it’s assigned to, is executed or compiled.

This two features are the basis for my way of implementing OOP concepts in Forth. The result is an easy to use OOP Toolkit with a syntax in the spirit of Forth, that is available as a loadable source code module.

The basic underlaying idea is, to use implicit context switching to assign methods to objects. An object, before executed or compiled, creates and activates its class specific search order to give access to its methods. A method found in this context, switches back to the default search order (or to another one), before it is executed or compiled itself. The context switching is done by context switching preludes.

I found that using OOP concepts with Forth can be very appealing, especially when combined with on demand loading of source code modules.

cspForth is distributed to let you gain your own OOP experience with Forth and to possibly trigger some discussion. It is released under the Artistic License. Please see the LICENSE file enclosed in the cspForth archive file.

cspForth is work in progress. Please keep in mind that it came into beeing as a test bed and not as a production system.

Installing cspForth

cspForth is distributed as cspForth-*.tar.gz or cspForth-*.zip archive file. Unpack the archive file in a directory of your choice. This will create a directory named csp4th-* with several subdirectories and files in it. We will call it the cspForth directory from now on.

In the cspForth directory you will find a file named csp4th. This is the cspForth executable. It should work out of the box with most current Linux distributions. If not, you have to build it from the source (see Building cspForth).

Note:

  • You can install cspForth on an USB-Stick for mobile use or to test it with a Linux Live System from CD or DVD.

Building cspForth

Coming soon.

Starting cspForth

The cspForth executable is called csp4th. You will find it in the cspForth directory. It can be invoked with or without options. Options may appear multiple times and in any order. When csp4th is invoked without any option it simply starts up, displays a startup message and waits for terminal input. If invoked with options, it starts up without a startup message and processes the options.

Syntax:

[/path/]csp4th [-e 'words ...'] [-n 'module name'] [-t 'module name'] [filename] [--]

Options:

-e 'words …' causes 'words ..' to be interpreted
-n 'module name' loads the file 'module/name.4th' from the cspForth source code library if it is not already loaded
-t 'module name' loads the file 'module/name.4th' from the cspForth source code library
filename includes the file 'filename'
stops processing of command line options

Before processing any option, cspForth interprets the hidden file .appnamerc in the users home directory, if it exits. appname is the name of the cspForth executable. The default is csp4th. This file may be used to configure cspForth during startup. It must contain valid Forth code. For security reasons, this file, if it exists, should only be editable by the user, it belongs to!

Notes:

  • As long as the cspForth directory is not in your path, you have to enter the relative or absolute file name to invoke the cspForth executable.
  • On a Linux System that does not support the /proc file system, you must enter the absolute file name to invoke the cspForth executable. Otherwise it will not startup properly. This may also be the case, if you try cspForth on an Ubuntu Live System.
  • cspForth does not support command line editing and command line history of itself. To get both you can call the cspForth executable via the rlwrap utility.
rlwrap [/path/]csp4th [-e 'words ...'] [-n 'module name'] [-t 'module name'] [filename] [--]

Using Modules

cspForth supports to create and use source code modules. A module is a text file holding valid Forth code. It has the file name extension .4th and is stored in one of two module directories. The names of this directories are hold in the variables dir(1) and dir(2). The defaults are user and share in the cspForth directory. You may change this ( with care ) to your needs. The default search order for modules is dir(1), dir(2).

You can write your own modules, give it the file name extension .4th and store it in one of the module directories. Use the share directory for modules that would be of interest for other users too.

Modules can be loaded once or multiple times.

needs module name

loads the module 'module name' and registers it as loaded. Any further 'needs module name' is then ignored.

take module name

loads the module module name without registering it and can be used to load a module again and again.

The module name is the file name of the related module relativ to the module directory but without the file name extension and with the file name separator characters '/' substituted with a space, e.g. needs String Array will load the file String/Array.4th from the modules directory dir(1)=<cspForth directory>/user or dir(2)='<cspForth directory>/share.

Note:

Text following a 'needs …' or 'take …' phrase in the same line, must be separated from the 'needs …' or 'take …' phrase by two or more spaces.

Getting Help

cspForth has a context sensitive help utility, implemented as a loadable module. The module can be loaded with needs help. It adds the word h to the dictionary.

To get help for a word <name> enter h <name> on the command line.

If <name> is found in the current search order, its glossary entry is displayed. Otherwise a global help file is searched for an entry for the word <name>. If one is found, it is displayed. If not, an error message is displayed.

Without entering a name you will get some help or info for the current context.

Using OOP Concepts

Loading the Modules

  needs oop ( -- )

Loads the OOP module, if it´s not yet loaded. It also loads all other modules required by the OOP module.

After loading the OOP module the system status is as follows :

needs oop  ??

  Stack: (0)
Current: oop
Context: oop  oop  forth  root
--------------------------------------------------------------------------------
 ok
  • the data stack is unchanged
  • the current compilation context is the vocabulary oop
  • the current search context is the vocabulary search order with the vocabulary oop on the top

Note: The vocabulary oop should be used as the “root” context for OOP-based definitions.

Not to Get Stuck

The three most important words of the OOP module are .. , ?? and ??? . They will help you not to get stuck in the class hierarchy and are available in any context.

  .. ( -- )

Switches back from a class, object or iface search order to the vocabulary search order, that was left, when the class, object or iface search order was activated.

Notes:

  • .. is the default prelude that is assigned to methods.
  • .. is a noop-word in the vocabulary search order.
  ?? ( -- )

Displays the current system status and the words of the top wordlist in the current search order.

  ??? ( -- )

Same as ?? but lists all words of all wordlists in the current search order, if it´s a class, object or iface search order.

Creating Classes

class <name> ( -- )

Creates a new class in the current compilation context, using the next word from the input stream as its name.

<name> definitions ( -- )

Makes the new class the current compilation context.

A new class may inherit from another class.

<class> inherit ( -- )

The object specification of a class may be an explicit memory request

self allot ( u -- )

or a list of instance variables ( references to objects of other classes )

<class1> this <name1> ( -- )
<class2> this <name2> ( -- )

or a mixture of both.

An object specification, when finalised, should be sealed explicitly,

self seal

although it is sealed implicitly when the first object is created.

The next step is to create methods to be applied to the objects of the class.

m: <name> ( i*x oid -- j*x )

starts a colon definition for a method. The Forth word m: has, compared with the word : (colon), the additional affect, to make the created word a context switching one, by assigning it the word .. as context switching prelude.

oid is the object identifier of the calling object. The programer is responsable to handle it in the methods definition. Sometimes this makes the code complicated, especially when using do … loop constructs. Then it's possible to get the oid out of the way by using an object stack. This gives a little runtime overhead but can make coding much easier.

To use the object stack, start the stack diagram of the method definition with (O. This compiles a word, that pushes the oid to the top of the object stack (TOOS), when the method is executed.

m: <name> (O i*x -- j*x ) 

Inside the method definition you can fetch the oid from the TOOS to the top of the data stack (TOS) with the word this. Its a context switching one. Its prelude is the word self, i.e. it activates the class search order, the method belongs to.

Note:

  • A method definition, like any colon definition, is terminated with the Forth word ; (semicolon). For methods, the word semicolon has an extra compile time semantics: If the word, preceeding the semicolon is a context switching one, switching to a class or iface context, then the default prelude of the method is overwritten with this word, i.e. the method will switch to that context.

A method can also be created as an alias of an existing Forth word. This is done with

method ' <name1> alias <name2>

or

m' <name1> alias <name2>

Creating Objects

To create an instance of a class use the word 'new'.

<class> new <object> ( -- )

This creates a variable with the class <class> as its prelude. The memory, needed for the object,is assigned to the variable. It was specified by the object specification of the class <class>. The object, when executed, returns the address of this memory on the top of the data stack (TOS). We call it the object identifier 'oid'.

Additional memory may be allocated with the 'init' method ( see the modules 'Buffer.4th' and 'String.4th' as examples ).

A new object should always be initialized, before it is used. The respective method must be defined by the programer and should always be called init.

<object> init ( i*x -- j*x )

The init method should at least initialize the objects memory.

Using Objects

An object can be any combination of data and methods. You can implement variables, datatypes, data structures, functional building blocks and whole applications. See the predefined classes and the examples in the module directories.

An object, when executed, returns its object identifier oid on the top of the data stack (TOS). It's the address of the objects data memory. In interpret or compile mode, an object gives access to the public instance variables and to the methods of the class it belongs to, to be executed or compiled.

Using Shared Libraries

cspForth supports to import functions from shared libraries. The (object oriented) library interface can be loaded with

  needs libs ( -- )

To import a function from a shared library one must first create and initialize a library object to get access to the library

           lib new <name> ( -- )    " <filename>" <name> init

  e.g.:    lib new libc ( -- )      " libc.so.6" libc init

A function from the library can then be imported with

         <name> import <function> ( input parameters -- output parameters )

  e.g.:    libc import time ( a|0 -- sec ) 

or

         <name> import <function> as <alias> ( input parameters -- output parameters )

  e.g.:    libc import time as os_time ( a|0 -- sec ) 

Note: The stack diagram is parsed to determine the number of input and output parameters.

For further information use the help utility in the contexts lib and libs.

Interfacing GTK+

Based on the OOP Module it was relatively easy to implement a GTK+-API for cspForth. GTK+ has an object oriented design, although it's written in C. It's class hierarchy can be mapped directly to a corresponding class hierarchy in cspForth.

GTK+ is a really big and powerful toolkit. So I have taken the approach to only implement a minimal subset of properties and methods per class. Further properties and methods can be added later, when needed for an application.

Classes are implemented as loadable source code modules, so one can always only load those classes, that are really needed for an application.

cspForth comes with a lot of predefined classes. Most of the simpler GTK+ Widgets are already implemented and others will follow (you are invited to contribute). A widget is created as an instance of its class and must be initialized explicitly with its init method, before any further use.

You will find examples for most of the predefined classes, that will help you figure out, how to use the GTK+ libraries. 1) To execute all this examples at one go, start cspForth and enter

take gtk examples

on the command line.

For the main class hierarchy you will already find glossary entries with the context sensitive help utility (after a module has been loaded). For most other classes, and for the examples, help texts are not yet written. Please have a look at the files in the modules directory dir(1)=<cspForth directory>/user. See the files hello.1.4th and GtkToplevel/example.x.4th to get started.

If you are familiar with the German Language you may want to read the article Widgets zum Anfassen - GUI-Skripting mit Forth und GTK+ in the Forth-Magazin Vierte Dimension 2/2008.

Saving cspForth

You can save the current state of a running cspForth system to an executable file. For further information see the help text for the word save.

Warning

cspForth is work in progress and that's much more the case for the GTK+ Interface. Details may change !

1)
I recently started to write a tutorial for the cspForth GTK+ API. You can find it here.
en/projects/csp4th/readme.txt · Last modified: 2013-06-06 21:26 by 127.0.0.1