A PicoLisp Tutorial

(c) Software Lab. Alexander Burger

About this document

This document demonstrates some aspects of the PicoLisp system in detail and example. For a general description of the PicoLisp kernel please look at the PicoLisp Reference.

This is not a Lisp tutorial, as it assumes some basic knowledge of programming, Lisp, and even PicoLisp. Please read these sections before coming back here: Introduction and The PicoLisp Machine. This tutorial concentrates on the specificities of PicoLisp, and its differences with other Lisp dialects.

Now let's start

If not stated otherwise, all examples assume that PicoLisp was started from a global installation (see Installation) from the shell prompt as

$ pil +

It loads the PicoLisp base system and the debugging environment, and waits for you to enter input lines at the interpreter prompt (:). You can terminate the interpreter and return to the shell at any time, by either hitting the Ctrl-D key, or by executing the function (bye).

Please note that special handling is done during character input. This one is incompatible with rlwrap for example but is more powerful. It supports vi-style command-line editing (typos fixes and history with ESC, h, j, k and l, or the arrow keys if supported by the terminal).

If you prefer to use Emacs, please use the picolisp-mode bundled in the "el/" directory (that is "@lib/el" for a local installation, or some system dependent directory for a global installation).

If you feel that you absolutely have to use an IDE, rlwrap or another input front-end, please create an empty "~/.pil/editor" file. This effectively disables the command line editor. Note that in this case, however, you will not have the TAB symbol completion feature available during command line editing.

Table of content

If you are new to PicoLisp, you might want to read the following sections in the given order, as some of them assume knowledge about previous ones. Otherwise just jump anywhere you are interested in.

Command Line Editing

PicoLisp permanently reads input from the current input channel (i.e. the console in interactive mode), evaluates it, and prints the result to the current output channel. This is called a "read-eval-print-loop" (REPL).


This is the default line editor, as it needs less system resources and works also on dumb terminals. It is similar to - though simpler than - the 'vi' edit modes of the 'korn' and 'bash' shells. For an analog 'emacs' style editor, please see below.

It is very helpful - though not absolutely necessary - when you know how to use the vi text editor.

To alleviate the task of manual line input, a command line editor is provided which is similar to (though much simpler than) the readline feature of the bash shell. Only a subset of the vi mode is supported, which is restricted to single-key commands (the "real" vi supports multi-key commands and the modification of most commands with count prefixes). It is loaded at startup in debug mode, you find its source in "lib/led.l".

You can enter lines in the normal way, correcting mistypes with the BACKSPACE key, and terminating them with the ENTER key. This is the Insert Mode.

If you hit ESC, you get into Command Mode. Now you can navigate horizontally in the current input line, or vertically in the history of previously entered lines, with key commands borrowed from the vi editor (only h, j, k and l, or the arrow keys if supported by the terminal). Note, however, that there is always only a single line visible.

Let's say you did some calculation

: (* (+ 2 3) (- 7 2))
-> 25

If you want to repeat a modified version of this command, using 8 instead of 7, you don't have to re-type the whole command, but type

Then you hit ENTER to execute the modified line. Instead of jumping to the 7 with the "find" command, you may also type l (move "right") repeatedly till you reach the correct position.

The key commands in the Command Mode are listed below. Some commands change the mode back to Insert Mode as indicated in parentheses. Deleting or changing a "word" take either the current atom (number or symbol), or a whole expression when the cursor is at a left parenthesis.


The following two key-combinations work both in Insert and Command Mode:

Besides these two keys, in Insert Mode only the following keys have a special meaning:


You can switch the command line editor to an 'emacs' style, if you call the function (em) (i.e. without arguments). A single call is enough. Alternatively, you could invoke PicoLisp at least once with the -em command line option

$ pil -em +

The style will be remembered in a file "~/.pil/editor", and used in all subsequent PicoLisp sessions.

To switch back to 'vi' style, call (vi), use the -vi command line option, or simply remove "~/.pil/editor".


Please take some time to experiment and to get used to command line editing. It will make life much easier in the future :-)


PicoLisp provides some functionality for inspecting pieces of data and code within the running system.

Basic tools

The really basic tools are of course available and their name alone is enough to know: print, size ...

But you will appreciate some more powerful tools like:

Inspect a symbol with show

The most commonly used tool is probably the show function. It takes a symbolic argument, and shows the symbol's name (if any), followed by its value, and then the contents of the property list on the following lines (assignment of such things to a symbol can be done with set, setq, and put).

: (setq A '(This is the value))  # Set the value of 'A'
-> (This is the value)
: (put 'A 'key1 'val1)           # Store property 'key1'
-> val1
: (put 'A 'key2 'val2)           # and 'key2'
-> val2
: (show 'A)                      # Now 'show' the symbol 'A'
A (This is the value)
   key2 val2
   key1 val1
-> A

show accepts an arbitrary number of arguments which are processed according to the rules of get, resulting in a symbol which is showed then.

: (put 'B 'a 'A)        # Put 'A' under the 'a'-property of 'B'
-> A
: (setq Lst '(A B C))   # Create a list with 'B' as second argument
-> (A B C)
: (show Lst 2 'a)       # Show the property 'a of the 2nd element of 'Lst'
A (This is the value)   # (which is 'A' again)
   key2 val2
   key1 val1
-> A

Inspect and edit with edit

Similar to show is edit. It takes an arbitrary number of symbolic arguments, writes them to a temporary file in a format similar to show, and starts the vim editor with that file.

: (edit 'A 'B)

The vim window will look like

A (This is the value)
key1 val1
key2 val2


a A  # (This is the value)


Now you can modify values or properties. You should not touch the parenthesized hyphens, as they serve as delimiters. If you position the cursor on the first character of a symbol name and type 'K' ("Keyword lookup"), the editor will be restarted with that symbol added to the editor window. 'Q' (for "Quit") will bring you back to the previous view.

And, pressing 'Q' at the first (or, top-level) view will return you to the REPL with any changes you made in your 'edit' session communicated back to the REPL session.

edit is also very useful to browse in a database. You can follow the links between objects with 'K', and even - e.g. for low-level repairs - modify the data (but only if you are really sure about what you are doing, and don't forget to commit when you are done).

Built-in pretty print with pp

The pretty-print function pp takes a symbol that has a function defined (or two symbols that specify message and class for a method definition), and displays that definition in a formatted and indented way.

: (pp 'pretty)
(de pretty (X N)
   (setq N (abs (space (or N 0))))
   (while (and (pair X) (== 'quote (car X)))
      (prin "'")
      (pop 'X) )
      (T (prtty0 X N)) ) )
-> pretty

The style is the same as we use in source files:

Inspect elements one by one with more

more is a simple tool that displays the elements of a list one by one. It stops after each element and waits for input. If you just hit ENTER, more continues with the next element, otherwise (usually I type a dot (.) followed by ENTER) it terminates.

: (more (1 2 3 4 5 6))
1                          # Hit ENTER
2.                         # Hit '.' and ENTER
-> T                       # stopped

Optionally more takes a function as a second argument and applies that function to each element (instead of the default print). Here, often show or pp (see below) is used.

: (more '(A B))            # Step through 'A' and 'B'
-> NIL
: (more '(A B) show)       # Step through 'A' and 'B' with 'show'
A (This is the value)      # showing 'A'
   key2 val2
   key1 val1
                           # Hit ENTER
B NIL                      # showing 'B'
   a A
-> NIL

Search through available symbols with what

The what function returns a list of all internal symbols in the system which match a given pattern (with '@' wildcard characters).

: (what "prin@")
-> (prin print prinl print> printsp println)

Search through values or properties of symbols with who

The function who returns "who contains that", i.e. a list of symbols that contain a given argument somewhere in their value or property list.

: (who 'print)
-> (query pretty pp msg more "edit" view show (print> . +Date) rules select
(print> . +relation))

A dotted pair indicates either a method definition or a property entry. So (print> . +relation) denotes the print> method of the +relation class.

who can be conveniently combined with more and pp:

: (more (who 'print) pp)
(de query ("Q" "Dbg")  # Pretty-print these functions one by one
   (use "R"
         (NIL (prove "Q" "Dbg"))
         (T (=T (setq "R" @)) T)
         (for X "R"
            (print (car X))
            (print '=)
            (print (cdr X))
            (flush) )
         (T (line)) ) ) )

(de pretty (X N)

The argument to who may also be a pattern list (see match):

: (who '(print @ (val @)))
-> (show)

: (more (who '(% @ 7)) pp)
(de day (Dat Lst)
   (when Dat
         (or Lst *DayFmt)
         (inc (% (inc Dat) 7)) ) ) )

(de _week (Dat)
   (/ (- Dat (% (inc Dat) 7)) 7) )

Find what classes can accept a given message with can

The function can returns a list which indicates which classes can accept a given message. Again, this list is suitable for iteration with pp:

: (can 'del>)                                   # Which classes accept 'del>' ?
-> ((del> . +List) (del> . +Entity) (del> . +relation))

: (more (can 'del>) pp)                         # Inspect the methods with 'pp'
(dm (del> . +List) (Obj Old Val)
   (and ((<> Old Val) (delete Val Old)) )

(dm (del> . +Entity) (Var Val)
         (has> (meta This Var) Val (get This Var)) )
      (let Old (get This Var)
            (meta This Var)
            (put This Var (del> (meta This Var) This Old @)) )
         (when (asoq Var (meta This 'Aux))
            (relAux This Var Old (cdr @)) )
         (upd> This Var Old) ) ) )

(dm (del> . +relation) (Obj Old Val)
   (and ((<> Old Val) Val) )

Inspect dependencies with dep

dep shows the dependencies in a class hierarchy. That is, for a given class it displays the tree of its (super)class(es) above it, and the tree of its subclasses below it.

To view the complete hierarchy of input fields, we start with the root class +relation:

: (dep '+relation)
-> +relation

If we are interested in +Link:

: (dep '+Link)
-> +Link

This says that +Link is a subclass of +relation, and has a single subclass (+Joint).

Defining Functions

Most of the time during programming is spent defining functions (or methods). In the following we will concentrate on functions, but most will be true for methods as well except for using dm instead of de.

Functions with no argument

The notorious "Hello world" function must be defined:

: (de hello ()
   (prinl "Hello world") )
-> hello

The () in the first line indicates a function without arguments. The body of the function is in the second line, consisting of a single statement. The last line is the return value of de, which here is the defined symbol. From now on we will omit the return values of examples when they are unimportant.

Now you can call this function this way:

: (hello)
Hello world

Functions with one argument

A function with an argument might be defined this way:

: (de hello (X)
   (prinl "Hello " X) )
# hello redefined
-> hello

PicoLisp informs you that you have just redefined the function. This might be a useful warning in case you forgot that a bound symbol with that name already existed.

: (hello "world")
Hello world

: (hello "Alex")
Hello Alex

Preventing arguments evaluation and variable number of arguments

Normally, PicoLisp evaluates the arguments before it passes them to a function:

: (hello (+ 1 2 3))
Hello 6

: (setq A 1  B 2)       # Set 'A' to 1 and 'B' to 2
-> 2
: (de foo (X Y)         # 'foo' returns the list of its arguments
   (list X Y) )
-> foo
: (foo A B)             # Now call 'foo' with 'A' and 'B'
-> (1 2)                # -> We get a list of 1 and 2, the values of 'A' and 'B'

In some cases you don't want that. For some functions (setq for example) it is better if the function gets all arguments unevaluated, and can decide for itself what to do with them.

For such cases you do not define the function with a list of parameters, but give it a single atomic parameter instead. PicoLisp will then bind all (unevaluated) arguments as a list to that parameter.

: (de foo X
   (list (car X) (cadr X)) )        # 'foo' lists the first two arguments

: (foo A B)                         # Now call it again
-> (A B)                            # -> We don't get '(1 2)', but '(A B)'

: (de foo X
   (list (car X) (eval (cadr X))) ) # Now evaluate only the second argument

: (foo A B)
-> (A 2)                            # -> We get '(A 2)'

Mixing evaluated arguments and variable number of unevaluated arguments

As a logical consequence, you can combine these principles. To define a function with 2 evaluated and an arbitrary number of unevaluated arguments:

: (de foo (X Y . Z)     # Evaluate only the first two args
   (list X Y Z) )

: (foo A B C D E)
-> (1 2 (C D E))        # -> Get the value of 'A' and 'B' and the remaining list

Variable number of evaluated arguments

More common, in fact, is the case where you want to pass an arbitrary number of evaluated arguments to a function. For that, PicoLisp recognizes the symbol @ as a single atomic parameter and remembers all evaluated arguments in an internal frame. This frame can then be accessed sequentially with the args, next, arg and rest functions.

: (de foo @
   (list (next) (next)) )     # Get the first two arguments

: (foo A B)
-> (1 2)

Again, this can be combined:

: (de foo (X Y . @)
   (list X Y (next) (next)) ) # 'X' and 'Y' are fixed arguments

: (foo A B (+ 3 4) (* 3 4))
-> (1 2 7 12)                 # All arguments are evaluated

These examples are not very useful, because the advantage of a variable number of arguments is not used. A function that prints all its evaluated numeric arguments, each on a line followed by its squared value:

: (de foo @
   (while (args)                            # Check if there are some args left
      (println (next) (* (arg) (arg))) ) )  # Call the last arg (next) returned

: (foo (+ 2 3) (- 7 1) 1234 (* 9 9))
5 25
6 36
1234 1522756
81 6561
-> 6561

This next example shows the behaviour of args and rest:

: (de foo @
   (while (args)
      (println (arg) (args) (rest)) ) )
: (foo 1 2 3)
1 T (2 3)
2 T (3)

Finally, it is possible to pass all these evaluated arguments to another function, using pass:

: (de foo @
   (pass println 9 8 7)       # First print all arguments preceded by 9, 8, 7
   (pass + 9 8 7) )           # Then add all these values

: (foo (+ 2 3) (- 7 1) 1234 (* 9 9))
9 8 7 5 6 1234 81             # Printing ...
-> 1350                       # Return the result

Anonymous functions without the lambda keyword

There's no distinction between code and data in PicoLisp, quote will do what you want (see also this FAQ entry).

: ((quote (X) (* X X)) 9)
-> 81

: (setq f '((X) (* X X)))  # This is equivalent to (de f (X) (* X X))
-> ((X) (* X X))
: f
-> ((X) (* X X))
: (f 3)
-> 9


There are two major ways to debug functions (and methods) at runtime: Tracing and single-stepping.

In this section we will use the REPL to explore the debugging facilities, but in the Scripting section, you will learn how to launch PicoLisp scripts with some selected functions debugged:

$ pil app/file1.l -"trace 'foo" -main -"debug 'bar" app/file2.l +


Tracing means letting functions of interest print their name and arguments when they are entered, and their name again and the return value when they are exited.

For demonstration, let's define the unavoidable factorial function (or just load the file "@doc/fun.l"):

(de fact (N)
   (if (=0 N)
      (* N (fact (dec N))) ) )

With trace we can put it in trace mode:

: (trace 'fact)
-> fact

Calling fact now will display its execution trace.

: (fact 3)
 fact : 3
  fact : 2
   fact : 1
    fact : 0
    fact = 1
   fact = 1
  fact = 2
 fact = 6
-> 6

As can be seen here, each level of function call will indent by an additional space. Upon function entry, the name is separated from the arguments with a colon (:), and upon function exit with an equals sign (=) from the return value.

trace works by modifying the function body, so generally it works only for functions defined as lists (lambda expressions, see Evaluation). Tracing a built-in function (SUBR) is possible, however, when it is a function that evaluates all its arguments.

So let's trace the functions =0 and *:

: (trace '=0)
-> =0
: (trace '*)
-> *

If we call fact again, we see the additional output:

: (fact 3)
 fact : 3
  =0 : 3
  =0 = NIL
  fact : 2
   =0 : 2
   =0 = NIL
   fact : 1
    =0 : 1
    =0 = NIL
    fact : 0
     =0 : 0
     =0 = 0
    fact = 1
    * : 1 1
    * = 1
   fact = 1
   * : 2 1
   * = 2
  fact = 2
  * : 3 2
  * = 6
 fact = 6
-> 6

To reset a function to its untraced state, call untrace:

: (untrace 'fact)
-> fact
: (untrace '=0)
-> =0
: (untrace '*)
-> *

or simply use mapc:

: (mapc untrace '(fact =0 *))
-> *


Single-stepping means to execute a function step by step, giving the programmer an opportunity to look more closely at what is happening. The function debug inserts a breakpoint into each top-level expression of a function. When the function is called, it stops at each breakpoint, displays the expression it is about to execute next (this expression is also stored into the global variable ^) and enters a read-eval-loop. The programmer can then

Thus, in the simplest case, single-stepping consists of just hitting ENTER repeatedly to step through the function.

To try it out, let's look at the stamp system function. You may need to have a look at

to understand this definition.

: (pp 'stamp)
(de stamp (Dat Tim)
   (and (=T Dat) (setq Dat (date T)))
   (default Dat (date) Tim (time T))
   (pack (dat$ Dat "-") " " (tim$ Tim T)) )
-> stamp

: (debug 'stamp)                       # Debug it
-> T
: (stamp)                              # Call it again
(and (=T Dat) (setq Dat (date T)))     # stopped at first expression
!                                      # ENTER
(default Dat (date) Tim (time T))      # second expression
!                                      # ENTER
(pack (dat$ Dat "-") " " (tim$ ...     # third expression
! Tim                                  # inspect 'Tim' variable
-> 41908
! (time Tim)                           # convert it
-> (11 38 28)
!                                      # ENTER
-> "2004-10-29 11:38:28"               # done, as there are only 3 expressions

Now we execute it again, but this time we want to look at what's happening inside the second expression.

: (stamp)                              # Call it again
(and (=T Dat) (setq Dat (date T)))
!                                      # ENTER
(default Dat (date) Tim (time T))
!                                      # ENTER
(pack (dat$ Dat "-") " " (tim$ ...     # here we want to look closer
! (d)                                  # debug this expression
-> T
!                                      # ENTER
(dat$ Dat "-")                         # stopped at first subexpression
! (e)                                  # evaluate it
-> "2004-10-29"
!                                      # ENTER
(tim$ Tim T)                           # stopped at second subexpression
! (e)                                  # evaluate it
-> "11:40:44"
!                                      # ENTER
-> "2004-10-29 11:40:44"               # done

The breakpoints still remain in the function body. We can see them when we pretty-print it:

: (pp 'stamp)
(de stamp (Dat Tim)
   (! and (=T Dat) (setq Dat (date T)))
   (! default Dat (date) Tim (time T))
   (! pack
      (! dat$ Dat "-")
      " "
      (! tim$ Tim T) ) )
-> stamp

To reset the function to its normal state, call unbug:

: (unbug 'stamp)

Often, you will not want to single-step a whole function. Just use edit (see above) to insert a single breakpoint (the exclamation mark followed by a space) as CAR of an expression, and run your program. Execution will then stop there as described above; you can inspect the environment and continue execution with ENTER when you are done.

Functional I/O

Input and output in PicoLisp is functional, in the sense that there are not variables assigned to file descriptors, which need then to be passed to I/O functions for reading, writing and closing. Instead, these functions operate on implicit input and output channels, which are created and maintained as dynamic environments.

Standard input and standard output are the default channels. Try reading a single expression:

: (read)
(a b c)        # Console input
-> (a b c)

To read from a file, we redirect the input with in. Note that comments and whitespace are automatically skipped by read:

: (in "@doc/fun.l" (read))
-> (de fact (N) (if (=0 N) 1 (* N (fact (dec N)))))

The skip function can also be used directly. To get the first non-white character in the file with char:

: (in "@doc/fun.l" (skip "#") (char))
-> "("

from searches through the input stream for given patterns. Typically, this is not done with Lisp source files (there are better ways), but for a simple example let's extract all items immediately following fact in the file,

: (in "@doc/fun.l" (while (from "fact ") (println (read))))
(dec N)

or the word following "(de " with till:

: (in "@doc/fun.l" (from "(de ") (till " " T))
-> "fact"

To read the contents of a whole file (or the rest of it starting from the current position):

: (in "f.l" (till NIL T))
-> "... file contents ..."

With line, a line of characters is read, either into a single transient symbol (the type used by PicoLisp for strings),

: (in "@doc/tut.html" (line T))
-> "<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://..."

or into a list of symbols (characters):

: (in "@doc/tut.html" (line))
-> ("<" "!" "D" "O" "C" "T" "Y" "P" "E" " " "H" "T" "M" "L" ...

line is typically used to read tabular data from a file. Additional arguments can split the line into fixed-width fields, as described in the reference manual. If, however, the data are of variable width, delimited by some special character, the split function can be used to extract the fields. A typical way to import the contents of such a file is:

(load "@lib/import.l")

(in '("bin/utf2" "importFile.txt")              # Pipe: Convert to UTF-8
   (until (eof)                                 # Process whole file
      (let L (split (line) "^I")                # TAB-delimited data
         ... use 'getStr', 'getNum' etc ...     # process them

Some more examples with echo:

(in "a"                                         # Copy the first 40 Bytes
   (out "b"                                     # from file "a" to file "b"
      (echo 40) ) )

(in "@doc/tut.html"                             # Show the HTTP-header
   (echo "<body>") )

(out "file.mac"                                 # Convert to Macintosh
   (in "file.txt"                               # from Unix or DOS format:
      (while (char)
            (case @
               ("^M" NIL)                       # ignore CR
               ("^J" "^M")                      # convert LF to CR
               (T @) ) ) ) ) )                  # otherwise no change

(out "c"                                        # Merge the contents of "a"
   (in "b"                                      # and "b" into "c"
      (in "a"
         (while (read)                          # Read an item from "a",
            (println @ (in -1 (read))) ) ) ) )  # print it with an item from "b"


There are two possibilities to get the PicoLisp interpreter into doing useful work: via command line arguments, or as a stand-alone script.

Command line arguments for the PicoLisp interpreter

The command line can specify either files for execution, or arbitrary Lisp expressions for direct evaluation (see Invocation): if an argument starts with a hyphen, it is evaluated, otherwise it is loaded as a file. A typical invocation might look like:

$ pil app/file1.l -main app/file2.l +

It loads the debugging environment, an application source file, calls the main function, and then loads another application source. In a typical development and debugging session, this line is often modified using the shell's history mechanisms, e.g. by inserting debugging statements:

$ pil app/file1.l -"trace 'foo" -main -"debug 'bar" app/file2.l +

Another convenience during debugging and testing is to put things into the command line (shell history) which would otherwise have to be done each time in the application's user interface:

$ pil app/file1.l -main app/file2.l -go -'login "name" "password"' +

The final production release of an application usually includes a shell script, which initializes the environment, does some bookkeeping and cleanup, and calls the application with a proper command line. It is no problem if the command line is long and complicated.

For small utility programs, however, this is overkill. Enter full PicoLisp scripts.

PicoLisp scripts

It is better to write a single executable file using the mechanisms of "interpreter files". If the first two characters in an executable file are "#!", the operating system kernel will pass this file to an interpreter program whose pathname is given in the first line (optionally followed by a single argument). This is fast and efficient, because the overhead of a subshell is avoided.

Let's assume you installed PicoLisp in the directory "/home/foo/picolisp/", and put links to the executable and the installation directory as:

$ ln -s /home/foo/picolisp /usr/lib/picolisp
$ ln -s /usr/lib/picolisp/bin/picolisp /usr/bin/picolisp
Then a simple hello-world script might look like:

#!/usr/bin/picolisp /usr/lib/picolisp/lib.l
(prinl "Hello world!")

If you write this into a text file, and use chmod to set it to "executable", it can be executed like any other command. Note that (because # is the comment character in PicoLisp) the first line will not be interpreted, and you can still use that file as a normal command line argument to PicoLisp (useful during debugging).

Grab command line arguments from PicoLisp scripts

The fact that a hyphen causes evaluation of command line arguments can be used to simulate something like command line options. The following script defines two functions a and f, and then calls (load T) to process the rest of the command line (which otherwise would be ignored because of the (bye) statement):

#!/usr/bin/picolisp /usr/lib/picolisp/lib.l

(de a ()
   (println '-a '-> (opt)) )

(de f ()
   (println '-f '-> (opt)) )

(load T)
(opt retrieves the next command line option)

Calling this script (let's say we named it "testOpts") gives:

$ ./testOpts -f abc
-f -> "abc"
$ ./testOpts -a xxx  -f yyy
-a -> "xxx"
-f -> "yyy"

We have to be aware of the fact, however, that the aggregation of arguments like

$ ./testOpts -axxx  -fyyy


$ ./testOpts -af yyy

cannot be achieved with this simple and general mechanism of command line processing.

Run scripts from arbitrary places on the host file system

Utilities are typically used outside the context of the PicoLisp environment. All examples above assumed that the current working directory is the PicoLisp installation directory, which is usually all right for applications developed in that environment. Command line file arguments like "app/file1.l" will be properly found.

To allow utilities to run in arbitrary places on the host file system, the concept of home directory substitution was introduced. The interpreter remembers internally at start-up the pathname of its first argument (usually "lib.l"), and substitutes any leading "@" character in subsequent file names with that pathname. Thus, to run the above example in some other place, simply write:

$ /home/foo/picolisp/pil @app/file1.l -main @app/file2.l +

that is, supply a full path name to the initial command (here 'pil'), or put it into your PATH variable, and prefix each file which has to be loaded from the PicoLisp home directory with a @ character. "Normal" files (not prefixed by @) will be opened or created relative to the current working directory as usual.

Stand-alone scripts will often want to load additional modules from the PicoLisp environment, beyond the "lib.l" we provided in the first line of the hello-world script. Typically, at least a call to

(load "@lib/misc.l")

(note the home directory substitution) will be included near the beginning of the script.

As a more complete example, here is a script which extracts the date, name and size of the latest official PicoLisp release version from the download web site, and prints it to standard output:

#!/usr/bin/picolisp /usr/lib/picolisp/lib.l

(load "@lib/misc.l" "@lib/http.l")

(use (@Date @Name @Size)
         '(@Date ~(chop " - <a href=\"") @Name "\"" ">"
             @Name ~(chop "</a> (") @Size )
         (client "software-lab.de" 80 "down.html"
            (from "Release Archive")
            (from "<li>")
            (till ",") ) )
      (prinl @Name)
      (prinl @Date " -- " @Size) ) )


Editing scripts

We recommend that you have a terminal window open, and try the examples by yourself. You may either type them in, directly to the PicoLisp interpreter, or edit a separate source file (e.g. "@doc/fun.l") in a second terminal window and load it into PicoLisp with

: (load "@doc/fun.l")

each time you have modified and saved it.

Editing scripts with vi

Once a function is loaded from a source file, you can call 'vim' directly on that function with

: (vi 'fact)

The function 'vi' opens the appropriate source file, and jumps to the right line where 'fact' is defined. When you modify it, you can simply call 'ld' to (re)load that source file

: (ld)

Objects and Classes

The PicoLisp object model is very simple, yet flexible and powerful. Objects as well as classes are both implemented as symbols. In fact, there is no formal difference between objects and classes; classes are more a conceptual design consideration in the head of the programmer than a physical reality.

Having said this, we declare that normally:

  1. A Class
  2. An Object

So the main difference between classes and objects is that the former ones usually are internal symbols. By convention, their names start with a '+'. Sometimes it makes sense, however, to create named objects (as global singletons, for example), or even anonymous classes.

Both classes and objects have a list in their value, consisting of method definitions (often empty for objects) and (super)class(es). And both classes and objects have local data in their property lists (often empty for classes). This implies, that any given object (as an instance of a class) may have private (object-local) methods defined.

It is rather difficult to contrive a simple OOP example. We constructed a hierarchy of geometric shapes, with a base class +Shape and two subclasses +Rectangle and +Circle.

The source code is included as "@doc/shape.l" in the PicoLisp distribution, so you don't have to type it in. Just load the file, or start it from the shell as:

$ pil @doc/shape.l +

Let's look at it piece by piece. Here's the base class:

(class +Shape)
# x y

(dm T (X Y)
   (=: x X)
   (=: y Y) )

(dm move> (DX DY)
   (inc (:: x) DX)
   (inc (:: y) DY) )

The first line '(class +Shape)' defines the symbol +Shape as a class without superclasses. The following method definitions will go to that class.

The comment '# x y' in the second line is just a convention, to indicate what instance variables (properties) that class uses. As PicoLisp is a dynamic language, a class can be extended at runtime with any number of properties, and there is nothing like a fixed object size or structure. This comment is a hint of what the programmer thinks to be essential and typical for that class. In the case of +Shape, x and y are the coordinates of the shape's origin.

Then we have two method definitions, using the keyword dm for "define method". The first method is special, in that its name is T. Each time a new object is created, and a method with that name is found in its class hierarchy, that method will be executed. Though this looks like a "constructor" in other programming languages, it should probably better be called "initializer". The T method of +Shape takes two arguments X and Y, and stores them in the object's property list.

The second method move> changes the object's origin by adding the offset values DX and DY to the object's origin.

Now to the first derived class:

(class +Rectangle +Shape)
# dx dy

(dm T (X Y DX DY)
   (super X Y)
   (=: dx DX)
   (=: dy DY) )

(dm area> ()
   (* (: dx) (: dy)) )

(dm perimeter> ()
   (* 2 (+ (: dx) (: dy))) )

(dm draw> ()
   (drawRect (: x) (: y) (: dx) (: dy)) )

+Rectangle is defined as a subclass of +Shape. The comment '# dx dy' indicates that +Rectangle has a width and a height in addition to the origin coordinates inherited from +Shape.

The T method passes the origin coordinates X and Y to the T method of the superclass (+Shape), then stores the width and height parameters into dx and dy.

Next we define the methods area> and perimeter> which do some obvious calculations, and a method draw> which is supposed to draw the shape on the screen by calling some hypothetical function drawRect.

Finally, we define a +Circle class in an analog way, postulating the hypothetical function drawCircle:

(class +Circle +Shape)
# r

(dm T (X Y R)
   (super X Y)
   (=: r R) )

(dm area> ()
   (*/ (: r) (: r) 31415927 10000000) )

(dm perimeter> ()
   (*/ 2 (: r) 31415927 10000000) )

(dm draw> ()
   (drawCircle (: x) (: y) (: r)) )

Now we can experiment with geometrical shapes. We create a rectangle at point (0,0) with a width of 30 and a height of 20, and keep it in the variable R:

: (setq R (new '(+Rectangle) 0 0 30 20))  # New rectangle
-> $134432824                             # returned anonymous symbol
: (show R)
$134432824 (+Rectangle)                   # Show the rectangle
   dy 20
   dx 30
   y 0
   x 0

We see that the symbol $134432824 has a list of classes '(+Rectangle)' in its value, and the coordinates, width and height in its property list.

Sending messages to that object

: (area> R)                               # Calculate area
-> 600
: (perimeter> R)                          # and perimeter
-> 100

will return the values for area and perimeter, respectively.

Then we move the object's origin:

: (move> R 10 5)                          # Move 10 right and 5 down
-> 5
: (show R)
$134432824 (+Rectangle)
   y 5                                    # Origin changed (0,0) -> (10,5)
   x 10
   dy 20
   dx 30

Though a method move> wasn't defined for the +Rectangle class, it is inherited from the +Shape superclass.

Similarly, we create and use a circle object:

: (setq C (new '(+Circle) 10 10 30))      # New circle
-> $134432607                             # returned anonymous symbol
: (show C)
$134432607 (+Circle)                      # Show the circle
   r 30
   y 10
   x 10
-> $134432607
: (area> C)                               # Calculate area
-> 2827
: (perimeter> C)                          # and perimeter
-> 188
: (move> C 10 5)                          # Move 10 right and 5 down
-> 15
: (show C)
$134432607 (+Circle)                      # Origin changed (10,10) -> (20,15)
   y 15
   x 20
   r 30

It is also easy to send messages to objects in a list:

: (mapcar 'area> (list R C))              # Get list of areas
-> (600 2827)
: (mapc
   '((Shape) (move> Shape 10 10))         # Move all 10 right and down
   (list R C) )
-> 25
: (show R)
$134431493 (+Rectangle)
   y 15
   x 20
   dy 20
   dx 30
-> $134431493
: (show C)
$134431523 (+Circle)
   y 25
   x 30
   r 30

Assume that we want to extend our shape system. From time to time, we need shapes that behave exactly like the ones above, but are tied to a fixed position. That is, they do not change their position even if they receive a move> message.

One solution would be to modify the move> method in the +Shape class to a no-operation. But this would require to duplicate the whole shape hierarchy (e.g. by defining +FixedShape, +FixedRectangle and +FixedCircle classes).

The PicoLisp Way is the use of Prefix Classes through multiple inheritance. It uses the fact that searching for method definitions is a depth-first, left-to-right search of the class tree. We define a prefix class:

: (class +Fixed)

(dm move> (DX DY))  # A do-nothing method

We can now create a fixed rectangle, and try to move it:

: (setq R (new '(+Fixed +Rectangle) 0 0 30 20))    # '+Fixed' prefix class
-> $134432881
: (move> R 10 5)                                   # Send 'move>' message
-> NIL
: (show R)
$134432881 (+Fixed +Rectangle)
   dy 20
   dx 30
   y 0                                             # Did not move!
   x 0

We see, prefix classes can surgically change the inheritance tree for selected objects or classes.

Alternatively, if fixed rectangles are needed often, it might make sense to define a new class +FixRect:

: (class +FixRect +Fixed +Rectangle)
-> +FixRect

and then use it directly:

: (setq R (new '(+FixRect) 0 0 30 20))
-> $13455710

Persistence (External Symbols)

PicoLisp has persistent objects built-in as a first class data type. With "first class" we mean not just the ability of being passed around, or returned from functions (that's a matter of course), but that they are a primary data type with their own interpreter tag bits. They are, in fact, a special type of symbolic atoms (called "External Symbols"), that happen to be read from pool file(s) when accessed, and written back automatically when modified.

In all other aspects they are normal symbols. They have a value, a property list and a name.

The name cannot be directly controlled by the programmer, as it is assigned when the symbol is created. It is an encoded index of the symbol's location in its database file. In its visual representation (output by the print functions and input by the read functions) it is surrounded by braces.

To make use of external symbols, you need to open a database first:

: (pool "test.db")

If a file with that name did not exist, it got created now. Also created at the same moment was {1}, the very first symbol in the file. This symbol is of great importance, and is handled especially by PicoLisp. Therefore a global constant *DB exists, which points to that symbol {1}, which should be used exclusively to access the symbol {1}, and which should never be modified by the programmer.

: *DB                   # The value of '*DB'
-> {1}                  # is '{1}'
: (show *DB)
{1} NIL                 # Value of '{1}' is NIL, property list empty

Now let's put something into the value and property list of {1}.

: (set *DB "Hello world")  # Set value of '{1}' to a transient symbol (string)
-> "Hello world"
: (put *DB 'a 1)           # Property 'a' to 1
-> 1
: (put *DB 'b 2)           # Property 'b' to 2
-> 2
: (show *DB)               # Now show the symbol '{1}'
{1} "Hello world"
   b 2
   a 1

Note that instead of '(set *DB "Hello world")', we might also have written '(setq {1} "Hello world")', and instead of '(put *DB 'a 1)' we might have written '(put '{1} 'a 1)'. This would have the same effect, but as a rule external symbols should never be be accessed literally in application programs, because the garbage collector might not be able to free these symbols and all symbols connected to them (and that might well be the whole database). It is all right, however, to access external symbols literally during interactive debugging.

Now we can create our first own external symbol. This can be done with new when a T argument is supplied:

: (new T)
-> {2}               # Got a new symbol

We store it in the database root {1}:

: (put *DB 'newSym '{2})   # Literal '{2}' (ok during debugging)
-> {2}
: (show *DB)
{1} "Hello world"
   newSym {2}              # '{2}' is now stored in '{1}'
   b 2
   a 1

Put some property value into '{2}'

: (put *DB 'newSym 'x 777) # Put 777 as 'x'-property of '{2}'
-> 777
: (show *DB 'newSym)       # Show '{2}' (indirectly)
{2} NIL
   x 777
-> {2}
: (show '{2})              # Show '{2}' (directly)
{2} NIL
   x 777

All modifications to - and creations of - external symbols done so far are not written to the database yet. We could call rollback (or simply exit PicoLisp) to undo all the changes. But as we want to keep them:

: (commit)           # Commit all changes
-> T
: (bye)              # Exit picolisp
$                    # back to the shell

So, the next time when ..

$ pil +                 # .. we start PicoLisp
: (pool "test.db")      # and open the database file,
-> T
: (show *DB)            # our two symbols are there again
{1} "Hello world"
   newSym {2}
   b 2
   a 1
-> {1}
: (show *DB 'newSym)
{2} NIL
   x 777
-> {2}

Database Programming

To a database, there is more than just persistence. PicoLisp includes an entity/relation class framework (see also Database) which allows a close mapping of the application data structure to the database.

We provided a simple yet complete database and GUI demo application in @doc/family.tgz and @doc/family64.tgz. Please unpack the first one if you use a 32-bit system, and the second one on a 64-bit system. Both contain the sources in @doc/family.l, and an initial database in the "family/" subdirectory.

To use it, please unpack it first in your current working directory, then start it up in the following way:

$ pil family.l -main +

This loads the source file, initializes the database by calling the main function, and prompts for user input.

The data model is small and simple. We define a class +Person and two subclasses +Man and +Woman.

(class +Person +Entity)

+Person is a subclass of the +Entity system class. Usually all objects in a database are of a direct or indirect subclass of +Entity. We can then define the relations to other data with the rel function.

(rel nm (+Need +Sn +Idx +String))      # Name

This defines the name property (nm) of a person. The second argument to rel is always a list of relation classes (subclasses of +relation), optionally followed by further arguments, causing relation daemon objects to be created and stored in the class definition. These daemon objects control the entity's behavior later at runtime.

Relation daemons are a kind of metadata, controlling the interactions between entities, and maintaining database integrity. Like other classes, relation classes can be extended and refined, and in combination with proper prefix classes a fine-grained description of the application's structure can be produced.

Besides primitive relation classes, like +Number, +String or +Date, there are

In the case of the person's name (nm) above, the relation object is of type (+Need +Sn +Idx +String). Thus, the name of each person in this demo database is a mandatory attribute (+Need), searchable with the soundex algorithm (+Sn) and a full index (+Idx) of type +String.

(rel pa (+Joint) kids (+Man))          # Father
(rel ma (+Joint) kids (+Woman))        # Mother
(rel mate (+Joint) mate (+Person))     # Partner

The attributes for father (pa), Mother (ma) and partner (mate) are all defined as +Joints. A +Joint is probably the most powerful relation mechanism in PicoLisp; it establishes a bidirectional link between two objects.

The above declarations say that the father (pa) attribute points to an object of type +Man, and is joined with that object's kids attribute (which is a list of joints back to all his children).

The consistency of +Joints is maintained automatically by the relation daemons. These become active whenever a value is stored to a person's pa, ma, mate or kids property.

For example, interesting things happen when a person's mate is changed to a new value. Then the mate property of the old mate's object is cleared (she has no mate after that). Now when the person pointed to by the new value already has a mate, then that mate's mate property gets cleared, and the happy new two mates now get their joints both set correctly.

The programmer doesn't have to care about all that. He just declares these relations as +Joints.

The last four attributes of person objects are just static data:

(rel job (+Ref +String))               # Occupation
(rel dat (+Ref +Date))                 # Date of birth
(rel fin (+Ref +Date))                 # Date of death
(rel txt (+String))                    # Info

They are all searchable via a non-unique index (+Ref). Date values in PicoLisp are just numbers, representing the day number (starting first of March of the year zero).

A method url> is defined:

(dm url> ()
   (list "!person" '*ID This) )

It is needed later in the GUI, to cause a click on a link to switch to that object.

The classes +Man and +Woman are subclasses of +Person:

(class +Man +Person)
(rel kids (+List +Joint) pa (+Person)) # Children

(class +Woman +Person)
(rel kids (+List +Joint) ma (+Person)) # Children

They inherit everything from +Person, except for the kids attribute. This attribute joins with the pa or ma attribute of the child, depending on the parent's gender.

That's the whole data model for our demo database application.

It is followed by a call to dbs ("database sizes"). This call is optional. If it is not present, the whole database will reside in a single file, with a block size of 256 bytes. If it is given, it should specify a list of items, each having a number in its CAR, and a list in its CDR. The CARs taken together will be passed later to pool, causing an individual database file with that size to be created. The CDRs tell what entity classes (if an item is a symbol) or index trees (if an item is a list with a class in its CAR and a list of relations in its CDR) should be placed into that file.

A handful of access functions is provided, that know about database relationships and thus allows higher-level access modes to the external symbols in a database.

For one thing, the B-Trees created and maintained by the index daemons can be used directly. Though this is rarely done in a typical application, they form the base mechanisms of other access modes and should be understood first.

The function tree returns the tree structure for a given relation. To iterate over the whole tree, the functions iter and scan can be used:

(iter (tree 'dat '+Person) '((P) (println (datStr (get P 'dat)) (get P 'nm))))
"1770-08-03" "Friedrich Wilhelm III"
"1776-03-10" "Luise Augusta of Mecklenburg-Strelitz"
"1797-03-22" "Wilhelm I"

They take a function as the first argument. It will be applied to all objects found in the tree (to show only a part of the tree, an optional begin- and end-value can be supplied), producing a simple kind of report.

More useful is collect; it returns a list of all objects that fall into a range of index values:

: (collect 'dat '+Person (date 1982 1 1) (date 1988 12 31))
-> ({2-M} {2-L} {2-E})

This returns all persons born between 1982 and 1988. Let's look at them with show:

: (more (collect 'dat '+Person (date 1982 1 1) (date 1988 12 31)) show)
{2-M} (+Man)
   nm "William"
   dat 724023
   ma {2-K}
   pa {2-J}
   job "Heir to the throne"

{2-L} (+Man)
   nm "Henry"
   dat 724840
   ma {2-K}
   pa {2-J}
   job "Prince"

{2-E} (+Woman)
   nm "Beatrice"
   dat 726263
   ma {2-D}
   job "Princess"
   pa {2-B}

If you are only interested in a certain attribute, e.g. the name, you can return it directly:

: (collect 'dat '+Person (date 1982 1 1) (date 1988 12 31) 'nm)
-> ("William" "Henry" "Beatrice")

To find a single object in the database, the function db is used:

: (db 'nm '+Person "Edward")
-> {2-;}

If the key is not unique, additional arguments may be supplied:

: (db 'nm '+Person "Edward"  'job "Prince"  'dat (date 1964 3 10))
-> {2-;}

The programmer must know which combination of keys will suffice to specify the object uniquely. The tree search is performed using the first value ("Edward"), while all other attributes are used for filtering. Later, in the Pilog section, we will show how more general (and possibly more efficient) searches can be performed.

User Interface (GUI) Programming

The only types of GUI supported by the PicoLisp application server framework is either dynamically generated (but static by nature) HTML, or an interactive XHTML/CSS framework with the optional use of JavaScript.

Before we explain the GUI of our demo database application, we present a minimal example for a plain HTML-GUI in @doc/hello.l. Start the application server as:

$ pil @lib/http.l  --server 8080 @doc/hello.l  -wait

Now point your browser to the address 'http://localhost:8080'. You should see a very simple HTML page. You can come back here with the browser's BACK button.

You can call the page repeatedly, or concurrently with many clients if you like. To terminate the server, you have to send it a TERM signal (e.g. 'killall pil'), or type the Ctrl-C key in the console window.

In our demo database application, a single function person is responsible for the whole GUI. Again, please look at @doc/family.l.

To start the database and the application server, call:

$ pil family.l -main -go +

As before, the database is opened with main. The function go is also defined in @doc/family.l:

(de go ()
   (server 8080 "!person") )

It starts the HTTP server listening on TCP port 8080 (we did a similar thing in our minimal GUI example above directly on the command line). Each connect to that port will cause the function person to be invoked.

Again, point your browser to the address 'http://localhost:8080'.

You should see a new browser window with an input form created by the function person. We provided an initial database in "family/[1-4]". You can navigate through it by clicking on the pencil icons besides the input fields.

The chart with the children data can be scrolled using the down (v) and up (^) buttons.

A click on the button "Select" below opens a search dialog. You can scroll through the chart as before. Again, a click on a pencil will jump to that person. You can abort the dialog with a click on the "Cancel"-button.

The search fields in the upper part of the dialog allow a conjunctive search. If you enter "Edward" in the "Name" field and click "Search", you'll see all persons having the string "Edward" in their name. If you also enter "Duke" in the "Occupation" field, the result list will reduce to only two entries.

To create a new person, press the "New Man" or "New Woman" button. A new empty form will be displayed. Please type a name into the first field, and perhaps also an occupation and birth date. Any change of contents should be followed by a press on the "Done" button, though any other button (also Scroll or Select-buttons) will also do.

To assign a father attribute, you can either type a name directly into the field (if that person already exists in the database and you know the exact spelling), or use the "Set"-button (->) to the left of that field to open the search dialog. If you type in the name directly, your input must exactly match upper and lower case.

Alternatively, you may create a new person and assign a child in the "Children" chart.

On the console where you started PicoLisp, there should a prompt have appeared just when the browser connected. You can debug the application interactively while it is running. For example, the global variable *Top always contains the top level GUI object:

: (show *Top)

To take a look at the first field on the form:

: (show *Top 'gui 1)

A production application would be started in a slightly different way:

$ pil family.l -main -go -wait

In that case, no debug prompt will appear. In both cases, however, two pil processes will be running now. One is the initial server process which will continue to run until it is killed. The other is a child process holding the state of the GUI in the browser. It will terminate some time after the browser is closed, or when (bye) or a Ctrl-D is entered at the PicoLisp prompt.

Now back to the explanation of the GUI function person:

(de person ()
      (html 0 (get (default *ID (val *DB)) 'nm) "@lib.css" NIL
         (form NIL
            (<h2> (<id> (: nm)))

For an in-depth explanation of that startup code, please refer to the guide to PicoLisp Application Development.

All components like fields and buttons are controlled by form. The function gui creates a single GUI component and takes the type (a list of classes) and a variable number of arguments depending on the needs of these classes.

   (gui '(+E/R +TextField) '(nm : home obj) 40 "Name")

This creates a +TextField with the label "Name" and a length of 40 characters. The +E/R (: Entity/Relation) prefix class connects that field to a database object, the nm attribute of a person in this case, so that the person's name is displayed in that text field, and any changes entered into that field are propagated to the database automatically.

   (gui '(+ClassField) '(: home obj) '(("Male" +Man) ("Female" +Woman)))

A +ClassField displays and changes the class of an object, in this case the person's sex from +Man to +Woman and vice versa.

As you see, there is no place where explicit accesses to the database have to be programmed, no select or update. This is all encapsulated in the GUI components, mainly in the +E/R prefix class. The above function person is fully functional as we present it and allows creation, modification and deletion of person objects in the database.

The two buttons on the bottom right generate simple reports:

The first one shows all contemporaries of the person that is currently displayed, i.e. all persons who did not die before, or were not born after that person. This is a typical PicoLisp report, in that in addition to the report's HTML page, a temporary file may be generated, suitable for download (and import into a spread sheet), and from which a PDF can be produced for print-out.

In PicoLisp, there is not a real difference between a plain HTML-GUI and a report. Again, the function html is used to generate the page.

The second report is much simpler. It produces a recursive structure of the family.

In both reports, links to the person objects are created which allow easy navigation through the database.

Pilog -- PicoLisp Prolog

This sections explains some cases of using Pilog in typical application programming, in combination with persistent objects and databases. Please refer to the Pilog section of the PicoLisp Reference for the basic usage of Pilog.

Again, we use our demo application @doc/family.l that was introduced in the Database Programming section.

Normally, Pilog is used either interactively to query the database during debugging, or in applications to generate export data and reports. In the following examples we use the interactive query front-end functions ? and select. An application will use goal and prove directly, or use convenience functions like pilog or solve.

All Pilog access to external symbols is done via the two predicates db/3 and select/3.

A predicate show/1 is pre-defined for debugging purposes (a simple glue to the Lisp-level function show, see Browsing). Searching with db/3 for all persons having the string "Edward" in their name:

: (? (db nm +Person "Edward" @P) (show @P))
{2-;} (+Man)
   nm "Edward"
   ma {2-:}
   pa {2-A}
   dat 717346
   job "Prince"
{2-1B} (+Man)
   nm "Albert Edward"
   kids ({2-1C} {2-1D} {2-1E} {2-1F} {2-1G} {2-1H} {2-1I} {2-g} {2-a})
   job "Prince"
   mate {2-f}
   fin 680370
   dat 664554
...               # more results

To search for all persons with "Edward" in their name who are married to somebody with occupation "Queen":

: (? (db nm +Person "Edward" @P) (val "Queen" @P mate job) (show @P))
{2-1B} (+Man)
   mate {2-f}
   nm "Albert Edward"
   kids ({2-1C} {2-1D} {2-1E} {2-1F} {2-1G} {2-1H} {2-1I} {2-g} {2-a})
   job "Prince"
   fin 680370
   dat 664554
-> NIL            # only one result

If you are interested in the names of "Albert Edward"'s children:

: (? (db nm +Person "Albert Edward" @P) (lst @K @P kids) (val @Kid @K nm))
 @P={2-1B} @K={2-1C} @Kid="Beatrice Mary Victoria"
 @P={2-1B} @K={2-1D} @Kid="Leopold George Duncan"
 @P={2-1B} @K={2-1E} @Kid="Arthur William Patrick"
 @P={2-1B} @K={2-1F} @Kid="Louise Caroline Alberta"
 @P={2-1B} @K={2-1G} @Kid="Helena Augusta Victoria"
 @P={2-1B} @K={2-1H} @Kid="Alfred Ernest Albert"
 @P={2-1B} @K={2-1I} @Kid="Alice Maud Mary"
 @P={2-1B} @K={2-g} @Kid="Victoria Adelaide Mary"
 @P={2-1B} @K={2-a} @Kid="Edward VII"
-> NIL

db/3 can do a direct index access only for a single attribute (nm of +Person above). To search for several criteria at the same time, select/3 has to be used:

: (?
   (select (@P)
      ((nm +Person "Edward") (nm +Person "Augusta" pa))  # Generator clauses
      (tolr "Edward" @P nm)                              # Filter clauses
      (tolr "Augusta" @P kids nm) )
   (show @P) )
{2-1B} (+Man)
   kids ({2-1C} {2-1D} {2-1E} {2-1F} {2-1G} {2-1H} {2-1I} {2-g} {2-a})
   mate {2-f}
   nm "Albert Edward"
   job "Prince"
   fin 680370
   dat 664554
-> NIL

select/3 takes a list of generator clauses which are used to retrieve objects from the database, and a number of normal Pilog filter clauses. In the example above the generators are

All persons generated are possible candidates for our selection. The nm index tree of +Person is traversed twice in parallel, optimizing the search in such a way that successful hits get higher priority in the search, depending on the filter clauses. The process will stop as soon as any one of the generators is exhausted. Note that this is different from the standard Prolog search algorithm.

The filter clauses in this example both use the pre-defined predicate tolr/3 for tolerant string matches (according either to the soundex algorithm (see the section Database Programming) or to substring matches), and filter objects that

A more typical and extensive example for the usage of select can be found in the qPerson function in @doc/family.l. It is used in the search dialog of the demo application, and searches for a person with the name, the parents' and partner's names, the occupation and a time range for the birth date. The relevant index trees in the database are searched (actually only those trees where the user entered a search key in the corresponding dialog field), and a logical AND of the search attributes is applied to the result.

For example, press the "Select" button, enter "Elizabeth" into the "Mother" search field and "Phil" in the "Partner" search field, meaning to look for all persons whose mother's name is like "Elizabeth" and whose partner's name is like "Phil". As a result, two persons ("Elizabeth II" and "Anne") will show up.

In principle, db/3 can be seen as a special case of select/3. The following two queries are equivalent:

: (? (db nm +Person "Edward" @P))
-> NIL
: (? (select (@P) ((nm +Person "Edward"))))
-> NIL

Poor Man's SQL


For convenience, a select Lisp glue function is provided as a front-end to the select predicate. Note that this function does not evaluate its arguments (it is intended for interactive use), and that it supports only a subset of the predicate's functionality. The syntax resembles SELECT in the SQL language, for example:

# SELECT * FROM Person
: (select +Person)  # Step through the whole database
{2-o} (+Man)
   nm "Adalbert Ferdinand Berengar Viktor of Prussia"
   dat 688253
   ma {2-j}
   pa {2-h}
   fin 711698

{2-1B} (+Man)
   nm "Albert Edward"
   dat 664554
   job "Prince"
   mate {2-f}
   kids ({2-1C} {2-1D} {2-1E} {2-1F} {2-1G} {2-1H} {2-1I} {2-g} {2-a})
   fin 680370

# SELECT * FROM Person WHERE nm LIKE "%Edward%"
: (select +Person nm "Edward")  # Show all Edwards
{2-;} (+Man)
   nm "Edward"
   dat 717346
   job "Prince"
   ma {2-:}
   pa {2-A}

{2-1B} (+Man)
   nm "Albert Edward"
   dat 664554
   job "Prince"
   kids ({2-1C} {2-1D} {2-1E} {2-1F} {2-1G} {2-1H} {2-1I} {2-g} {2-a})
   mate {2-f}
   fin 680370

# SELECT nm, dat FROM Person WHERE nm LIKE "%Edward%"
: (select nm dat +Person nm "Edward")
"Edward" "1964-03-10" {2-;}
"Albert Edward" "1819-08-26" {2-1B}
"George Edward" NIL {2-R}
"Edward Augustus Hanover" NIL {2-1K}

# SELECT dat, fin, p1.nm, p2.nm
#    FROM Person p1, Person p2
#    WHERE p1.nm LIKE "%Edward%"
#    AND p1.job LIKE "King%"
#    AND p1.mate = p2.mate  -- Actually, in a SQL model we'd need
#                           -- another table here for the join
: (select dat fin nm (mate nm) +Person nm "Edward" job "King")
"1894-06-23" "1972-05-28" "Edward VIII" "Wallace Simpson" {2-T}
"1841-11-09" NIL "Edward VII" "Alexandra of Denmark" {2-a}
-> NIL


In addition (just to stay with the SQL terminology ;-), there is also an update function. It is a front-end to the set!> and put!> transaction methods, and should be used when single objects in the database have to be modified by hand.

In principle, it would also be possible to use the edit function to modify a database object. This is not recommended, however, because edit does not know about relations to other objects (like Links, Joints and index trees) and may easily cause database corruption.

In the most general case, the value of a property in a database object is changed with the put!> method. Let's look at "Edward" from the previous examples:

: (show '{2-;})
{2R} (+Man)
   job "Prince"
   nm "Edward"
   dat 717346
   ma {2-:}
   pa {20A}
-> {2-;}

We might change the name to "Johnny" with put!>:

: (put!> '{2-;} 'nm "Johnny")
-> "Johnny"

However, an easier and less error-prone prone way - especially when more than one property has to be changed - is using update. It presents the value (the list of classes) and then each property on its own line, allowing the user to change it with the command line editor.

Just hitting ENTER will leave that property unchanged. To modify it, you'll typically hit ESC to get into command mode, and move the cursor to the point of change.

For properties with nested list structures (+List +Bag), update will recurse into the data structure.

: (update '{2-;})
{2-;} (+Man)      # ENTER
nm "Johnny"       # Modified the name to "Johnny"
ma {2-:}          # ENTER
pa {2-A}          # ENTER
dat 1960-03-10    # Modified the year from "1964" to "1960"
job "Prince"      # ENTER
-> {2-;}

All changes are committed immediately, observing the rules of database synchronization so that any another user looking at the same object will have his GUI updated correctly.

To abort update, hit Ctrl-X.

If only a single property has to be changed, update can be called directly for that property:

: (update '{2-;} 'nm)
{2-;} nm "Edward"


[knuth73] Donald E. Knuth: ``The Art of Computer Programming'', Vol.3, Addison-Wesley, 1973, p. 392