Contents
- Introduction
- newLISP script files
- Specifying command line options in script files
- Scripts as pipes
- File filters
- Writing software in modules
- Structuring an application
- More than one context per file
- Simple context objects and the default function
- Packaging data with contexts
- Passing context objects by reference
- Local variables
- Locals in looping functions
- Local symbols using let, letn and local
- Unused parameters as local symbols
- Default values for unfilled variables
- Using args as local substitute
- Named parameters using bind
- Walking through lists
- Recursion or not?
- Generators
- Speed up with memoization
- Walking a tree
- Walking a directory tree
- Creating, accessing, modifying and searching lists
- push and pop
- Accessing lists
- Selecting more than one element
- Filtering and differencing lists
- Changing list elements
- The new element as a function of the replaced
- Search and replace in simple lists
- Search and replace in nested lists
- Passing lists by reference using the default functor
- Program flow
- Loops
- Blocks
- Branching
- Fuzzy flow
- Change flow with catch and throw
- Leave loops with a break condition
- Change flow with and or or
- Error handling
- newLISP errors
- User defined errors
- Error event handlers
- Catching errors
- Functions as data
- Manipulate functions after definition
- Mapping and applying functions
- Function currying: functions making functions
- Text processing
- Regular expressions
- Scanning text
- Appending strings
- Growing strings in place
- Rearranging strings
- Modifying strings
- Dictionaries, hashes
- Dictionaries for hash-like key → value access
- Saving and loading dictionaries
- Updating dictionaries entries
- TCP/IP client server communications
- Client - server TCP/IP - open connection
- Client - server TCP/IP - closed connection
- UDP communications
- Open communications with UDP
- Closed transaction oriented UDP
- UDP multi-cast communications
- Non-blocking communications
- Using net-select
- Using net-peek
- Controlling other applications
- Controlling an application with exec
- Communications via STD I/O pipes
- Communications via TCP/IP
- Communicate via named FIFO
- Communicate via UDP
- Launching applications blocking
- Shell execution
- Execution capturing std-out in a list
- Execution feeding std-in
- Processes, semaphores and shared memory
- Multiprocessing and the Cilk API
- Starting concurrent processes
- Watching progress
- Invoking spawn recursively
- Event driven notification
- Message exchange between processes
- Collecting synchronous messages
- Collecting asynchronous messages
- Implementing message timeouts
- Evaluating messages
- Databases, lookup tables
- Association lists
- Nested associations
- Updating nested associations
- In-memory database with keyed access
- Distributed computing and server mode
- Setting up a newLISP server node
- Evaluating multiple expressions remotely
- Transferring files to and from remote nodes
- Loading and saving code from and to remote nodes
- Local domain UNIX sockets
- HTTPD-only mode, newLISP as a web-server
- CGI processing in HTTP mode
- Media types in HTTP mode
- Environment variables set
- Extending newLISP
- Compile a shared library on Linux/UNIX/Mac OS X
- Compile a DLL on Win32
- Using data structures in imported libraries
- Unevenly aligned data structures
- Passing parameters in library calls
- Extracting return values from library calls
- Writing library wrappers
- Appendix
- GNU Free Documentation License
When programming in newLISP certain functions and usage patterns occur repeatedly. For some problems an optimal way to solve them evolves over time. The following chapters present example code and explanations for the solution of specific problems when programming in newLISP.
Some content is overlapping with material covered in the newLISP Users Manual and Reference or presented here with a different slant.
Only a subset of newLISP's total function repertoire is used here. Some functions demonstrated have additional calling patterns or applications not mentioned on these pages.
This collection of patterns and solutions is a work in progress. Over time material will be added or existing material improved.
On Linux/UNIX put the following in the first line of the script/program file:
#!/usr/bin/newlisp
specifying a bigger stack:
#!/usr/bin/newlisp -s 100000
or
#!/usr/bin/newlisp -s100000
Operating systems shells behave differently when parsing the first line and extract parameters. newLISP takes both, attached or detached parameters. Put the following lines in small script to test the behavior of the underlying OS and platform. The script changes the stack size allocated to 100,000 and limits LISP cell memory to about 10 M bytes.
#!/usr/bin/newlisp -s 100000 -m 10
(println (main-args))
(println (sys-info))
A typical output executing the script from the system shell would be:
./arg-test
("/usr/bin/newlisp" "-s" "100000" "-m" "10" "./arg-test")
(308 655360 299 2 0 100000 8410 2)
Note that few programs in newLISP need a bigger stack configured, most programs run on the internal default of 2048. Each stack position takes an average of 80 bytes. Other options are available to start newLISP. See the Users Manual for details.
The following examples shows how a file can be piped into a newLISP script.
#!/usr/bin/newlisp
#
# uppercase - demo filter script as pipe
#
# usage:
# ./uppercase < file-spec
#
# example:
# ./uppercase < my-text
#
#
(while (read-line) (println (upper-case (current-line))))
(exit)
The file will be printed to std-out translated to uppercase.
The following script works like a Unix grep utility iterating through files and filtering each line in a file using a regular expression pattern.
#!/usr/bin/newlisp
#
# nlgrep - grep utility on newLISP
#
# usage:
# ./nlgrep "regex-pattern" file-spec
#
# file spec can contain globbing characters
#
# example:
# ./nlgrep "this|that" *.c
#
# will print all line containing 'this' or 'that' in *.c files
#
(dolist (file-name (3 (main-args)))
(set 'file (open file-name "read"))
(println "file ---> " file-name)
(while (read-line file)
(if (find (main-args 2) (current-line) 0)
(write-line)))
(close file))
(exit)
The expression:
(3 (main-args))
is a short form of writing:
(rest (rest (rest (main-args))))
It returns a list of all the filenames. This form of specifying indexes for rest is called implicit indexing. See the Users Manual for implicit indexing with other functions. The expression (main-args 2) extracts the 3rd argument from the command line containing the regular expression pattern.
When writing bigger applications or when several programmers are working on the same code base it is necessary to divide the code base into modules. Modules in newLISP are implemented using contexts, which are namespaces. Namespaces allow lexical isolation between modules. Variables of the same name in one module cannot clash with variables of the same name in another module.
Typically modules will be organized in one context per file. One file module may contain database access routines.
; database.lsp
;
(context 'db)
(define (update x y z)
...
)
(define (erase x y z)
...
)
Another module may contain various utilities
; auxiliary.lsp
;
(context 'aux)
(define (getval a b)
...
)
Typically there will be one MAIN module loading and controlling all others:
; application.lsp
;
(load "auxiliary.lsp")
(load "database.lsp")
(define (run)
(db:update ....)
(aux:putval ...)
...
...
)
(run)
When using more than one context per file, each context section should be closed with a (context MAIN) statement:
; myapp.lsp
;
(context 'A)
(define (foo ...) ...)
(context MAIN)
(context 'B)
(define (bar ...) ...)
(context MAIN)
(define (main-func)
(A:foo ...)
(B:bar ...)
)
Note that in the namespace statements for contexts A and B the context names are quoted because they are newly created, while MAIN already exists when newLISP starts up and can stay unquoted, although quoting it would not represent a problem.
To save the line closing the context with MAIN, the following technique can be used:
; myapp.lsp
;
(context 'A)
(define (foo ...) ...)
(context 'MAIN:B)
(define (bar ...) ...)
(context 'MAIN)
(define (main-func)
(A:foo ...)
(B:bar ...)
)
The line (context 'MAIN:B) switches back to MAIN then opens the new context B.
A function in a context may have the same name as the host context itself. This function has special characteristics:
(context 'foo)
(define (foo:foo a b c)
...
)
The function foo:foo is called the default function, because when using the context name foo like a function it will default to foo:foo:
(foo x y z)
; same as
(foo:foo x y z)
The default function make is possible to write functions which look like normal functions but carry their own lexical namespace. We can use this to write functions which keep state:
(context 'generator)
(set 'acc 0)
(define (generator:generator)
(inc acc))
(context MAIN)
(generator) → 1
(generator) → 2
(generator) → 3
The following is a more complex example for a function generating a fibonacci sequence:
(define (fibo:fibo)
(if (not fibo:mem) (set 'fibo:mem '(0 1)))
(last (push (+ (fibo:mem -1) (fibo:mem -2)) fibo:mem -1)))
(fibo) → 1
(fibo) → 2
(fibo) → 3
(fibo) → 5
(fibo) → 8
...
This example also shows how a default function is defined on the fly without the need of explicit context statements. As an alternative the function could also have been written creating the context explicitly:
(context 'fibo)
(define (fibo:fibo)
(if (not mem) (set 'mem '(0 1)))
(last (push (+ (mem -1) (mem -2)) mem -1)))
(context MAIN)
(fibo) → 1
(fibo) → 2
(fibo) → 3
(fibo) → 5
(fibo) → 8
while the first form is shorter, the second form is more readable.
The previous examples already presented functions packaged with data in a namespace. In the generator example the acc variable kept state. In the fibo example the variable mem kept a growing list. In both cases functions and data are living together in a namespace. The following example shows how a namespace just holds data only in a default functor:
(set 'db:db '(a "b" (c d) 1 2 3 x y z))
Just like we used the default function to refer to fibo and generator we can refer to the list in db:db by only using db. This will work in all situations where we do list indexing:
(db 0) → a
(db 1) → "b"
(db 2 1) → d
(db -1) → z
(db -3) → x
(3 db) → (1 2 3 x y z)
(2 1 db) → ((c d))
(-6 2 db) → (1 2)
The default functor when used as an argument in a user defined function, is passed by reference. That means that no copy of the list or string is passed but a reference to the original contents. This is useful when handling large lists or strings:
(define (update data idx expr)
(if (not (or (lambda? expr) (primitive? expr)))
(setf (data idx) expr)
(setf (data idx) (expr $it))))
(update db 0 99) → a
db:db → (99 "b" (c d) 1 2 3 x y z)
(update db 1 upper-case) → "b"
db:db → (99 "B" (c d) 1 2 3 x y z)
(update db 4 (fn (x) (mul 1.1 x))) →
db:db → (99 "B" (c d) 1 2.2 3 x y z)
The data in db:db is passed via the update function parameter data, which now holds a reference to the context db. The expr parameter passed is checked if a built-in function, operator or a user defined lambda expression and than works on $it, the anaphoric system variable containing the old content referenced by (data idx).
Whenever a function in newLISP asks for a string or list in a parameter, a default functor can be passed by its context symbol. Another example:
(define (pop-last data)
(pop data -1))
(pop-last db) → z
db:db → (99 "B" (c d) 1 2.2 3 x y)
The function update is also a good example how to pass operators or functions as function argument (upper-case working on $it). Read more about this in chapter 9. Functions as data.
All looping functions like doargs, dolist, dostring, dotimes, dotree and for, use local variables. During loop execution the variable takes different values, but after leaving the looping function the variable regains its old value. let, define, and lambda expressions are another method for making variables local:
let is the usual way in LISP to declare symbols as local to a block.
(define (sum-sq a b)
(let ((x (* a a)) (y (* b b)))
(+ x y)))
(sum-sq 3 4) → 25
; alternative syntax
(define (sum-sq a b)
(let (x (* a a) y (* b b))
(+ x y)))
; using local
(define (sum-sq a b)
(local (x y)
(set 'x (* a a))
(set 'y (* b b))
(+ x y)))
The variables x and y are initialized, then the expression (+ x y) is evaluated. The let form is just an optimized version and syntactic convenience for writing:
((lambda (sym1 [sym2 ...]) exp-body ) exp-init1 [ exp-init2 ...])
When initializing several parameters, a nested let, letn can be used to reference previously initialized variables in subsequent initializer expressions:
(letn ((x 1) (y (+ x 1)))
(list x y)) → (1 2)
The function local works like a let but initializing all variables to nil.
In newLISP all parameters in user defined functions are optional. Unused parameters are filled with nil and local to the dynamic scope of the function. Defining a user function with more parameters than required is a convenient method to create local variable symbols:
(define (sum-sq a b , x y)
(set 'x (* a a))
(set 'y (* b b))
(+ x y))
The comma is not a special syntax feature but only a visual helper to separate normal parameters from local variable symbols.
In the definition of a function default values can be specified:
(define (foo (a 1) (b 2))
(list a b))
(foo) → (1 2)
(foo 3) → (3 2)
(foo 3 4) → (3 4)
Using the args function no parameter symbols need to be used at all and args returns a list of all parameters passed but not taken by declared parameters:
(define (foo)
(args))
(foo 1 2 3) → (1 2 3)
(define (foo a b)
(args))
(foo 1 2 3 4 5) → (3 4 5)
The second example shows how args only contains the list of arguments not bound by the variable symbols a and b.
Indices can be used to access members of the (args) list:
(define (foo)
(+ (args 0) (args 1)))
(foo 3 4) → 7
bind can also be used together with local and args to write functions with named parameters:
(define-macro (foo)
(local (len width height)
(bind (args) true)
(println "len:" len " width:" width " height:" height)
))
> (foo (width (+ 15 5)) (height 30) (len 10))
len:10 width:20 height:30
> (foo (width (+ 15 5)) (height (* width 2)) (len 10))
len:10 width:20 height:40
>
Arguments are evaluated incrementally in the local environment and assigned to the local variables.
Although recursion is a powerful feature to express many algorithms in a readable form, they are also inefficient in some instances. newLISP has many iterative constructs and high level functions like flat or the built-in XML functions, which use recursion internally. In many cases this makes defining a recursive algorithm not necessary.
Some times a non-recursive solution can be much faster and lighter on system resources.
;; classic recursion
;; slow and resource hungry
(define (fib n)
(if (< n 2) 1
(+ (fib (- n 1))
(fib (- n 2)))))
The recursive solution is slow because of frequent calling overhead incurs. The recursive solution uses also a lot of memory for holding intermediate results and frequently redundant resuls.
;; iteration
;; fast and also returns the whole list
(define (fibo n , f)
(set 'f '(1 0))
(dotimes (i n)
(push (+ (f 0) (f 1)) f)) )
The iterative solution is fast and uses very little memory.
A generator is a function which keeps and changed state between invocations and returns a new value on each call:
;; generator uses a namespace to keep state
;; the default funcion (fibo) gets called repeatedly
;;
;; (fibo) → 1
;; (fibo) → 2
;; (fibo) → 3, 5, 8, 13, 21 ...
;;
;; fibo:mem → (0 1 1 2 3 5 8 13 21 ...)
(define (fibo:fibo)
(if (not fibo:mem) (set 'fibo:mem '(0 1)))
(last (push (+ (fibo:mem -2) (fibo:mem -1)) fibo:mem -1)))
Namespaces in newLISP together with default functions are a convenient method to package a function with local static variables. mem is used in the fibo name space to keep the Fibonacci sequence.
A memoizing function caches results for faster retrieval when called with the same parameters again. The following function makes a memoizing function from any built-in or user defined function with an arbitrary number of arguments:
;; speed up a recursive function using memoization
(define-macro (memoize mem-func func)
(set (sym mem-func mem-func)
(letex (f func c mem-func)
(lambda ()
(or (context c (string (args)))
(context c (string (args)) (apply f (args))))))))
(define (fibo n)
(if (< n 2) 1
(+ (fibo (- n 1))
(fibo (- n 2)))))
(memoize fibo-m fibo)
(time (fibo-m 25)) → 148
(time (fibo-m 25)) → 0
The function creates a context and default function for the original function with a new name and stores all results in symbols in the same context.
When memoizing recursive functions, include the the raw lambda specification of the function so recursive calls are memoized too:
(memoize fibo
(lambda (n)
(if(< n 2) 1
(+ (fibo (- n 1))
(fibo (- n 2))))))
(time (fibo 100)) → 1
(fibo 80) → 37889062373143906
The fibo function in the last example would take hours to calculate without memoization. The memoized version takes only about a milli-second for an argument of 100.
Tree walks are a typical pattern in traditional LISP and in newLISP as well for walking through a nested list. But many times a tree walk is only used to iterate through all elements of an existing tree or nested list. In this case the built-in flat function is much faster than using recursion:
(set 'L '(a b c (d e (f g) h i) j k))
;; classic car/cdr and recursion
;;
(define (walk-tree tree)
(cond ((= tree '()) true)
((atom? (first tree))
(println (first tree))
(walk-tree (rest tree)))
(true
(walk-tree (first tree))
(walk-tree (rest tree)))))
;; classic recursion
;; 3 times faster
;;
(define (walk-tree tree)
(dolist (elmnt tree)
(if (list? elmnt)
(walk-tree elmnt)
(println elmnt))))
(walk-tree L) →
a
b
c
d
e
...
Using the built-in flat in newLISP a nested list can be transformed int a flat list. Now the list can be processed with a dolist or map:
;; fast and short using 'flat'
;; 30 times faster with map
;;
(map println (flat L))
(dolist (item (flat L)) (println item)
Walking a directory tree is a task where recursion works well:
; walks a disk directory and prints all path-file names
;
(define (show-tree dir)
(if (directory dir)
(dolist (nde (directory dir))
(if (and (directory? (append dir "/" nde))
(!= nde ".") (!= nde ".."))
(show-tree (append dir "/" nde))
(println (append dir "/" nde))))))
In this example recursion is the only solution, because the entire nested list off files is not available when the function is called but gets created recursively during function execution.
newLISP has facilities for multidimensional indexing into nested lists. There are destructive functions like push, pop, setf, set-ref, set-ref-all, sort and reverse and many others for non-destructive operations, like nth, ref, ref-all, first, last and rest etc.. Many of the list functions in newLISP also work on strings.
Note that any list or string index in newLISP can be negative starting with -1 from the right side of a list:
(set 'L '(a b c d))
(L -1) → d
(L -2) → c
(-3 2 L) → (b c)
(set 'S "abcd")
(S -1) → d
(S -2) → c
(-3 2 S) → "bc")
To add to a list use push, to eliminate an element from a list use pop. Both functions are destructive, changing the contents of a list:
(set 'L '(b c d e f))
(push 'a L) → (a b c d e f)
(push 'g L -1) ;; push at the end with negative index
(pop L) ; pop first a
(pop L -1) ; pop last g
(pop L -2) ; pop second to last e
(pop L 1) ; pop second c
L → (b d f)
; multidimensional push / pop
(set 'L '(a b (c d (e f) g) h i))
(push 'x L 2 1) → (a b (c x d (e f) g) h i)
L → (a b (c x d (e f) g) h i)
(pop L 2 1) → x
Pushing to the end of a list repeatedly is optimized in newLISP and as fast as pushing in front of a list.
When pushing an element with index vector V it can be popped with the same index vector V:
(set 'L '(a b (c d (e f) g) h i))
(set 'V '(2 1))
(push 'x L V)
L → (a b (c x d (e f) g) h i))
(ref 'x L) → (2 1) ; search for a nested member
(pop L V) → 'v
Multiple indexes can be specified to access elements in a nested list structure:
(set 'L '(a b (c d (e f) g) h i))
; old syntax only for one index
(nth 2 L) → (c d (e f) g)
; use new syntax for multiple indices
(nth '(2 2 1) L) → f
(nth '(2 2) L) → (e f)
; vector indexing
(set 'vec '(2 2))
(nth vec L) → (e f)
; implicit indexing
(L 2 2 1) → f
(L 2 2) → (e f)
; implicit indexing with vector
(set 'vec '(2 2 1))
(L vec) → f
Implicit indexing shown in the last example cane make code more readable. Indexes before a list select subsections of a list, which in turn are always lists.
Implicit indexing is also available for rest and slice:
(rest '(a b c d e)) → (b c d e)
(rest (rest '(a b c d e) → (c d e)
; same as
(1 '(a b c d e)) → (b c d e)
(2 '(a b c d e)) → (c d e)
; negative indices
(-2 '(a b c d e)) → (d e)
; slicing
(2 2 '(a b c d e f g)) → (c d)
(-5 3 '(a b c d e f g)) → (c d e)
Sometimes more than one element must be selected from a list. This is done using select:
;; pick several elements from a list
(set 'L '(a b c d e f g))
(select L 1 2 4 -1) → (b c e g)
;; indices can be delivered in an index vector:
(set 'vec '(1 2 4 -1))
(select L vec) → (b c e g)
The selecting process can re-arrange or double elements at the same time:
(select L 2 2 1 1) → (c c b b)
Sometimes lists need to be filterer for a specific conditions applied to its elements:
(filter (fn(x) (< 5 x)) '(1 6 3 7 8)) → (6 7 8)
(filter symbol? '(a b 3 c 4 "hello" g)) → (a b c g)
(difference '(1 3 2 5 5 7) '(3 7)) → (1 2 5)
The first example could be rewritten shorter as follows:
(filter (curry < 5) '(1 6 3 7 8))
The curry function makes a one-argument function out of a two argument function:
(curry < 5) → (lambda (_x) (< 5 _x))
Using curry a function taking two arguments quickly is converted into a predicate taking one argument.
setf can be used to change a list element by referencing it with either nth or assoc:
; modify a list
(set 'L '(a b (c d (e f) g) h i))
(setf (L 2 2 1) 'x) → x
L → (a b (c d (e x) g) h i)
; modify an association list
(set 'A '((a 1) (b 2) (c 3)))
(setf (assoc 'b A) '(b 222)) → (b 222)
A → ((a 1) (b 222) (c 3))
The internal anaphoric system variable $it holds the old list element. This can be used to configure the new one:
(set 'L '(0 0 0))
(setf (L 1) (+ $it 1)) → 1 ; the new value
(setf (L 1) (+ $it 1)) → 2
(setf (L 1) (+ $it 1)) → 4
L → '(0 3 0)
Replace, which can also be used on strings, can search for and replace multiple elements in a list at once. Together with match and unify complex search patterns can be specified. Like with setf, the replacement expression can use the old element contents to form the replacement.
(set 'aList '(a b c d e a b c d))
(replace 'b aList 'B) → (a B c d e a B c d)
The function replace can take a comparison function for picking list elements:
; replace all numbers where 10 < number
(set 'L '(1 4 22 5 6 89 2 3 24))
(replace 10 L 10 <) → (1 4 10 5 6 10 2 3 10)
Using the built-in functions match and unify more complex selection criteria can be defined:
; replace only sublists starting with 'mary'
(set 'AL '((john 5 6 4) (mary 3 4 7) (bob 4 2 7 9) (jane 3)))
(replace '(mary *) AL (list 'mary (apply + (rest $it))) match)
→ ((john 5 6 4) (mary 14) (bob 4 2 7 9) (jane 3))
; make sum in all expressions
(set 'AL '((john 5 6 4) (mary 3 4 7) (bob 4 2 7 9) (jane 3)))
(replace '(*) AL (list ($it 0) (apply + (rest $it))) match)
→ ((john 15) (mary 14) (bob 22) (jane 3))
$0 → 4 ; replacements made
; change only sublists where both elements are the same
(replace '(X X) '((3 10) (2 5) (4 4) (6 7) (8 8)) (list ($it 0) 'double ($it 1)) unify)
→ ((3 10) (2 5) (4 double 4) (6 7) (8 double 8))
$0 → 2 ; replacements made
During replacements $0 and the anaphoric system variable $it contain the current found expression.
After a replacement statement is executed the newLISP system variable $0 contains the number of replacements made.
Sometimes lists are nested, e.g. the SXML results fropm parsing XML. The functions ref-set, set-ref and set-ref-all can be used to find and replace a single or all element in a nested list and replace it or all.
(set 'data '((monday (apples 20 30) (oranges 2 4 9)) (tuesday (apples 5) (oranges 32 1))))
(set-ref 'monday data tuesday)
→ ((tuesday (apples 20 30) (oranges 2 4 9)) (tuesday (apples 5) (oranges 32 1)))
The function set-ref-all does a set-ref multiple times, replacing all found occurrences of and element.
(set 'data '((monday (apples 20 30) (oranges 2 4 9)) (tuesday (apples 5) (oranges 32 1))))
(set-ref-all 'apples data "Apples")
→ ((monday ("Apples" 20 30) (oranges 2 4 9)) (tuesday ("Apples" 5) (oranges 32 1)))
Like find, replace, ref and ref-all, more complex searches can be expressed using match or unify:
(set 'data '((monday (apples 20 30) (oranges 2 4 9)) (tuesday (apples 5) (oranges 32 1))))
(set-ref-all '(oranges *) data (list (first $0) (apply + (rest $it))) match)
→ ((monday (apples 20 30) (oranges 15)) (tuesday (apples 5) (oranges 33)))
The last exampls shows how $0 can be used to access the old list element in the updating expression. In this case the numbers for oranges records have been summed up. Instead of $0 the anaphoric system variable $it can also be used.
Sometimes a larger list (more than a few hundred elements) must passed to a userdefined function for changing elements in it. Normally newLISP passes all parameters to user-defined functions by value. The following snipped shows a technique that can be used to pass a bigger list or string object by reference:
(set 'data:data '(a b c d e f g h))
(define (change db i value)
(setf (db i) value))
(change data 3 999) → d
data:data → '(a b c 999 d e f g h)
In this example the list is encapsulated in a context named data holding a variable data with the same name.
Whenever a function in newLISP looks for a string or list parameter, a context can be passed, which then will be interpreted as the default functor.
Many built-in functions when returning a list or array or an element belonging to a list or array referenced by a symbol, return a reference to the list – not a copy. This can be used to nest built-in functions when modifying a list:
(set 'L '(r w j s r b))
(pop (sort L)) → b
L → (j r r s w)
Program flow in newLISP is mostly functional but it also has looping and branching constructs and a catch and throw to break out of the normal flow.
Looping expressions as a whole behave like a function or block returning the last expression evaluated.
Most of the traditional looping patterns are supported. Whenever there is a looping variable, it is local in scope to the loop, behaving according the rules of dynamic scoping inside the current name-space or context:
; loop a number of times
; i goes from 0 to N - 1
(dotimes (i N)
....
)
; demonstrate locality of i
(dotimes (i 3)
(print i ":")
(dotimes (i 3) (print i))
(println))
→ ; will output
0:012
1:012
2:012
; loop through a list
; takes the value of each element in aList
(dolist (e aList)
...
)
; loop through a sring
; takes the ASCII or UTF-8 value of each character in aString
(dostring (e aString)
...
)
; loop through the symbols of a context in
; alphabetical order of the symbol name
(dotree (s CTX)
...
)
; loop from to with optional step size
; i goes from init to <= N inclusive with step size step
; Note that the sign in step is irrelevant, N can be greater
; or less then init.
(for (i init N step)
...
)
; loop while a condition is true
; first test condition then perform body
(while condition
...
)
; loop while a condition is false
; first test condition then perform body
(until condition
....
)
; loop while a condition is true
; first perform body then test
; body is performed at least once
(do-while condition
...
)
; loop while a condition is false
; first perform body then test
; body is performed at least once
(do-until condition
...
)
Note that the looping functions dolist, dotimes and for can also take a break condition as an additional argument. When the break condition evaluates to true the loop finishes:
(dolist (x '(a b c d e f g) (= x 'e))
(print x))
→ ; will output
abcd
Blocks are collections of s-expressions evaluated sequentially. All looping constructs may have expression blocks after the condition expression as a body.
Blocks can also be constructed by enclosing them in a begin expression:
(begin
s-exp1
s-exp2
.....
s-expN)
Looping constructs do not need to use an explicit begin after the looping conditions. begin is mostly used to block expressions in if and cond statements.
The functions and, or, let, letn and local can also be used to form blocks and do not require begin for blocking statements.
(if condition true-expr false-expr)
;or when no false clause is present
(if condition true-expr)
;or unary if for (filter if '(...))
(if condition)
;; more then one statement in the true or false
;; part must be blocked with (begin ...)
(if (= x Y)
(begin
(some-func x)
(some-func y)
(begin
(do-this x y)
(do-that x y)))
; the when form can take several statements without
; using a (begin ...) block
(when condition
exp-1
exp-2
...
)
; if-not works like (if (not ...) ...)
(if-not true-expr false-expr)
; unless works like (when (not ...) ...)
(unless condition
exp-1
exp-2
...
)
Depending on condition the exp-true or exp-false part is evaluated and returned.
More than one condition exp-true pair can occur in an if expression, making it look like a cond:
(if condition-1 exp-true-1
condition-2 exp-true-2
...
condition-n exp-true-n
false-expr)
The first exp-true-i of which the condition-i is not nil is evaluated and returned or the exp-false if none of the condition-i is true.
cond works like the multiple condition form of if but each part of condition-i exp-true-i must be braced in parenthesis:
(cond
(condition-1 exp-true-1 )
(condition-2 exp-true-2 )
...
(condition-n exp-true-n )
(true exp-true))
Using amb the program flow can be regulated in a probabilistic fashion:
(amb
exp-1
exp-2
...
exp-n)
One of the alternative expressions exp-1 to exp-n is evaluated with a probability of p = 1/n and the result is returned from the amb expression.
Any loop or other expression block can be enclosed in a catch expression. The moment a throw expression is evaluated, the whole catch expression returns the value of the throw expression.
(catch
(dotimes (i 10)
(if (= i 5) (throw "The End"))
(print i " ")))
; will output
0 1 2 3 4
; and the return value of the catch expression will be
→ "The End"
Several catch may be nested.
Loops built using dotimes, dolist or for can specify a break condition for leaving the loop early:
(dotimes (x 10 (> (* x x) 9))
(println x))
→
0
1
2
3
(dolist (i '(a b c nil d e) (not i))
(println i))
→
a
b
c
Similar to programming in the Prolog language, the logical and and or can be used to control program flow depending on the outcome of expressions logically connected:
(and
expr-1
expr-2
...
expr-n)
Expressions are evaluated sequentially until one expr-i evaluates to nil or the empty list () or until all expr-i are exhausted. The last expression evaluated is the return value of the whole and expression.
(or
expr-1
expr-2
...
expr-n)
Expressions are evaluated sequentially until one expr-i evaluates to not nil and not () or until all expr-i are exhausted. The last expression evaluated is the return value of the whole or expression.
Several conditions during evaluation of a newLISP expression can cause error exceptions. For a complete list of errors see the Appendix in the newLISP Reference Manual.
newLISP errors are caused by wrong syntax, of function invocation, not supplying the right amount of parameters, or supplying parameters with the wrong data type. Or trying to evaluate non existing functions.
; examples of newLISP errors
;
(foo foo) → invalid function : (foo foo)
(+ "hello") → value expected in function + : "hello"
User errors are error exceptions thrown using the function throw-error:
; user defined error
;
(define (double x)
(if (= x 99) (throw-error "illegal number"))
(+ x x))
(double 8) → 16
(double 10) → 20
(double 99)
→
user error : illegal number
called from user defined function double
newLISP and user defined errors can be caught using the function error-event to define an event handler.
; define an error event handler
;
(define (MyHandler)
(println "An error #" (error-number) " has occurred"))
(error-event 'MyHandler)
(foo) → An error #23 has occurred
A fine grainier, more specific error exception handling can be achieved using a special syntax of the function catch.
(define (double x)
(if (= x 99) (throw-error "illegal number"))
(+ x x))
catch with a second parameter can be used to catch both, system and user defined errors:
(catch (double 8) 'result) → true
result → 16
(catch (double 99) 'result) → nil
(print result)
→
user error : illegal number
called from user defined function double
(catch (double "hi") 'result) → nil
(print result)
→
value expected in function + : x
called from user defined function double
The catch expression returns true when no error exception occurred and the result of the expression is found in the symbol result specified as a second parameter.
If an error exception occurs, it is caught and the catch clause returns nil. In this case the symbol result contains the error message.
Some errors originating at operating system level are not caught by newLISP, but can be inspected using the function sys-error. For example the failure to open a file could have different causes:
;; trying to open a nonexistent file
(open "blahbla" "r") → nil
(sys-error) → (2 "No such file or directory")
;; to clear errno specify 0
(sys-error 0) → (0 "Unknown error: 0")
Numbers returned may be different on different UNIX platforms. Consult the file /usr/include/sys/errno.h on your platform.
(define (double x) (+ x x))
→ (lambda (x) (+ x x))
(first double) → (x)
(last double) → (+ x x)
;; make a ''fuzzy'' double
(setf (nth 1 double) '(mul (normal x (div x 10)) 2))
(double 10) → 20.31445313
(double 10) → 19.60351563
lambda in newLISP is not an operator or symbol, but rather a special s-expression or list attribute:
(first double) → (x) ; not lambda
The lambda attribute of an s-expression is right-associative in append:
(append (lambda) '((x) (+ x x))) → (lambda (x) (+ x x))
; or shorter
(append (fn) '((x) (+ x x))) → (lambda (x) (+ x x))
(set 'double (append (lambda) '((x) (+ x x)))
(double 10) → 20
and left-associative when using cons:
(cons '(x) (lambda) → (lambda (x))
Lambda expressions in newLISP never loose their first class object property.
The word lambda can be abbreviated as fn, which is convenient when mapping or applying functions to make the expression more readable and shorter to type.
Functions or operators can be applied to a list of data at once and all results are returned in a list
(define (double (x) (+ x x))
(map double '(1 2 3 4 5)) → (2 4 6 8 10)
Functions can be applied to parameters occurring in a list:
(apply + (sequence 1 10)) → 55
Here and expression is passed as a parameter:
; macro expansion using expand
(define (raise-to power)
(expand (fn (base) (pow base power)) 'power))
; or as an alternative using letex
(define (raise-to power)
(letex (p power) (fn (base) (pow base p))))
(define square (raise-to 2))
(define cube (raise-to 3))
(square 5) → 25
(cube 5) → 125
The built-in function curry can be used to make a function taking one argument from a function taking two arguments.
(define add-one (curry add 1)) → (lambda () (add 1 ($args 0)))
(define by-ten (curry mul 10)) → (lambda () (mul 10 ($args 0)))
(add-one 5) → 6
(by-ten 1.23) → 12.3
Note that the curried parameter is always the first parameter of the original function.
Regular expression in newLISP can be used together with a variety of functions:
| directory | Return a list of files, can use a regex patterns for filtering. |
| ends-with | Test if a string ends with a key string or pattern. |
| find | Is used to find the position / offset of a pattern. |
| find-all | Assemble a list of all patterns found. |
| parse | Break a string into token at patterns found between tokens. |
| regex | Find patterns and lists all sub patterns with offset and length found. |
| replace | Replace found patterns with a user defined function, which can take the as input the patterns itself. |
| search | Search for a pattern in a file. |
| starts-with | Test if a string starts with a key string or pattern. |
The functions find, regex, replace and search store pattern matching results in the system variable $0 to $15. See the newLISP Users Manual for details.
The following paragraphs show frequently used algorithms for scanning and tokenizing text.
replace together with a regular expression pattern can be used to scan text. The pattern in this case describes the tokens scanned for. As each token is found it is pushed on a list. The work is done in the replacement expression part of replace. This example saves all files linked on a web page:
; tokenize using replace with regular expressions
(set 'page (get-url "http://www.nodep.nl/newlisp/index.html"))
(replace {href="(http://.*lsp)"} page (first (push $1 links)) 0)
(dolist (link links)
(set 'file (last (parse link "/")))
(write-file file (get-url link))
(println "->" file))
Curly braces {,} are used in the regular expression pattern to avoid escaping the quotes " or other character with special meaning in regular expressions.
The following alternative technique is even shorter. The find-all function puts all pattern matching strings into a list:
(set 'links (find-all {href="(http://.*lsp)"} page))
(dolist (link links)
(set 'file (last (parse link "/")))
(write-file file (get-url link))
(println "->" file))
In an additional expressions find-all can be directed to do additional work with subexpressions found:
(find-all {(new)(lisp)} "newLISPisNEWLISP" (append $2 $1) 1)
→ ("LISPnew" "LISPNEW")
In the last example find-all appends the sub expressions found in reverse order before returning them in the result list.
Another technique for tokenizing text uses parse. While with replace and find-all the regular expression defined the token, when using parse the regex pattern describes the space between the tokens:
; tokenize using parse
(set 'str "1 2,3,4 5, 6 7 8")
(parse str {,\ *|\ +,*} 0)
→ ("1" "2" "3" "4" "5" "6" "7" "8")
Without the curly braces {,} in the parse pattern the backslashes would need to be doubled. Note that there is a space after the backslashes.
When appending strings append and join can be used to form a new string:
(set 'lstr (map string (rand 1000 100)))
→ ("976" "329" ... "692" "425")
;; the wrong slowest way
(set 'bigStr "")
(dolist (s lstr)
(set 'bigStr (append bigStr s)))
;; smarter way - 50 times faster
;;
(apply append lstr)
Sometimes strings are not readily available in a list like in the above examples. In this case push can be used to push strings on a list while they get produced. The list then can be used as an argument for join, making the fastest method for putting strings together from existing pieces:
;; smartest way - 300 times faster
;; join an existing list of strings
;;
(join lstr) → "97632936869242555543 ...."
;; join can specify a string between the elements
;; to be joined
(join lstr "-") → "976-329-368-692-425-555-43 ...."
Very often the best method is to grow a string in place. The functions write-buffer, write-line and push can be used not only to write to file handles or push on lists but also to write/append to existing strings:
;; smartest way - 150 times faster
;; grow a string in place
;;
;; using write-buffer
(set 'bigStr "")
(dolist (s lstr) (write-buffer bigStr s))
;; using push
(set 'bigStr "")
(dolist (s lstr) (push s bigStr -1))
The function select for selecting elements from lists can also be used to select and re-arrange characters from strings:
(set 'str "eilnpsw")
(select str '(3 0 -1 2 1 -2 -3)) → "newlisp"
; alternative syntax
(select str 3 0 -1 2 1 -2 -3) → "newlisp"
The second syntax is useful when indexes are specified not as constants but occur as variables.
newLISP has a variety of functions, which can destructively change a string:
replace replace all occurrences of a string or string pattern with a string setf replace a chracter in a string with one or more characters push pop Insert or extract one or more characters at a specific position
replace can also be used to remove all occurrences of string or string pattern when specifying an empty string "" as replacement.
When indexing strings with either nth or implicit indexing, the string is addressed at character rather than byte boundaries to work correctly on UTF-8 enabled versions of newLISP.
newLISP has functions to create and manipulate symbols using the functions sym and a special syntax of the function context. In older versions of newLISP these functions where used to program hash-like creation and access of key-value pairs. Now a shorter more convenient method is available using the un-initialized default functor of a context (namespace):
(define Myhash:Myhash) ; extabish the namespace and default functor
As an alternative to above methods the predefined namespace and default functor Tree can be used to instantiate a new one:
(new Tree 'Myhash)
Both methods produce the same result.
A default functor is the symbol with the same name as the namespace (context) it belongs too. If this default functor symbol does not contain anything except nil, it works like a hash function:
(Myhash "var" 123) ; create and set variable/value pair
(Myhash "var") → 123 ; retrieve value
(Myhash "foo" "hello")
(Myhash "bar" '(q w e r t y))
(Myhash "!#@$" '(a b c))
Setting a hash symbol to nil will effectively erase it:
(Myhash "bar" nil)
The key can be any string, symbol clashes with built-in newLISP symbols are avoided by newLISP prepending and underscore character to all key strings internally. The value can be any string, number or any other other newLISP s-expression.
The Myhash namespace can be transformed in an association list:
(Myhash) → (("!#@$" (a b c)) ("foo" "hello") ("var" 123))
Or the raw contents of Myhash can be shown using the symbols function:
(symbols Myhash) → (Myhash:Myhash Myhash:_!#@$ Myhash:_foo Myhash:_var)
Dictionaries can be built converting an existing association list:
(set 'aList '(("one" 1) ("two" 2) ("three")))
(Myhash aList)
(Myhash) → (("!#@$" (a b c)) ("foo" "hello") ("one" 1) ("three" nil) ("two" 2) ("var" 123))
The dictionary can be easily saved to a file by serializing the namespace Myhash:
(save "Myhash.lsp" 'Myhash)
The whole namespace is saved to the file Myhash.lsp and can be reloaded into newLISP at a later time:
(load "Myhash")
Note that hashes create contexts similar to the bayes-train function. All string keys are prepended with an underscore and then transfomed into a symbol. This means that namespaces created using bayes-train can be used like hashes to retrieve words and their statistics. See the bayes-train function in the manual for more detail.
In this pattern the server keeps the connection open until the client closes the connection, then the server loops into a new listen:
;;;;;;;;;;;;;;;;;;;;;; the server
; maximum bytes to receive
(constant 'max-bytes 1024)
(if (not (set 'listen (net-listen 123)))
(print (net-error)))
(while (not (net-error))
(set 'connection (net-accept listen)) ;; blocking here
(while (not (net-error))
(net-receive connection message-from-client max-bytes)
.... process message from client ...
.... configure message to client ...
(net-send connection message-to-client)) )
;;;;;;;;;;;;;;;;;;;;; the client
; client connect
(if (not (set 'connection (net-connect "host.com" 123)))
(println (net-error)))
; maximum bytes to receive
(constant 'max-bytes 1024)
; message send-receive loop
(while (not (net-error))
.... configure message to server ...
(net-send connection message-to-server)
(net-receive connection message-from-server max-bytes)
.... process message-from-server ...
)
In this pattern the server closes the connection after each transaction exchange of messages.
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; server
(while (not (net-error))
(set 'connection (net-accept listen)) ;; blocking here
(net-receive connection message-from-client max-bytes)
.... process message from client ...
.... configure message to client ...
(net-send connection message-to-client)
(close connection))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; client
(if (not (set 'connection (net-connect "host.com" 123)))
(println (net-error)))
; maximum bytes to receive
(constant 'max-bytes 1024)
.... configure message to server ...
(net-send connection message-to-server)
(net-receive connection message-from-server max-bytes)
.... process message-from-server ...
There are many different ways to set up a client/server connection, see also the examples in the newLISP manual.
They are fast and need less setup than TCP/IP and offer multi casting. UDP is also less reliable because the protocol does less checking, i.e. of correct packet sequence or if all packets are received. This is normally no problem when not working on the Internet but in a well controlled local network or when doing machine control. A simple more specific protocol could be made part of the message.
In this example the server keeps the connection open. UDP communications with net-listen, net-receive-from and net-send-to can block on receiving.
Note that both, the client and server use net-listen with the "udp" option. In this case net-listen is only used for binding the socket to the address, it is not used for listening for an connection. The server could receive messages from several clients. The net-send-to function extracts the target address from the message received.
;;;;;;;;;;;;;;;;;; server
(set 'socket (net-listen 10001 "" "udp"))
(if socket (println "server listening on port " 10001)
(println (net-error)))
(while (not (net-error))
(set 'msg (net-receive-from socket 255))
(println "->" msg)
(net-send-to (nth 1 msg) (nth 2 msg)
(upper-case (first msg)) socket))
;;;;;;;;;;;;;;;;;; client
(set 'socket (net-listen 10002 "" "udp"))
(if (not socket) (println (net-error)))
(while (not (net-error))
(print "->")
(net-send-to "127.0.0.1" 10001 (read-line) socket)
(net-receive socket buff 255)
(println "→" buff))
This form is some times used for controlling hardware or equipment. No setup is required, just one function for sending, another one for receiving:
;;;;;;;;;;;;;;;;;; server
;; wait for data gram with maximum 20 bytes
(net-receive-udp 1001 20)
;; or
(net-receive-udp 1001 20 5000000) ;; wait for max 5 seconds
;;;;;;;;;;;;;;;;;; client
(net-send-udp "host.com" 1001 "Hello")
Win32 and Linux show different behavior when sending less or more bytes then specified on the receiving end.
In this scheme the server subscribes to one of a range of multi cast addresses using the net-listen function.
;; example server
(net-listen 4096 "226.0.0.1" "multi") → 5
(net-receive-from 5 20)
;; example client I
(net-connect "226.0.0.1" 4096 "multi") → 3
(net-send 3 "hello")
;; example client II
(net-connect "" 4096 "multi") → 3
(net-send-to "226.0.0.1" 4096 "hello" 3)
The connection in the example is blocking on net-receive but could be de-blocked using net-select or net-peek
In all previous patterns the client blocks when in receive. The net-select call can be used to unblock communications:
;; optionally poll for arriving data with 100ms timeout
(while (not (net-select connection "r" 100000))
(do-something-while-waiting ...))
(net-receive...)
connection can be a single number for a connection socket or a list of numbers to wait on various sockets.
(while ( = (net-peek aSock) 0)
(do-something-while-waiting ...))
(net-receive...)
This method is only suited for short exchanges, executing one command and receiving the output.
(exec "ls *.c") → ("." ".." "util.c" "example.ls")
The exec function opens a process pipe for the Unix command-line utility ls and collects each line of STDOUT into a list of strings.
Most following examples use process to launch an application. This function returns immediately after launching the other application and does not block.
In all of the following patterns the server is not independent but controlled by the client, which launches the server and then communicates via a line oriented protocol:
→ launch server
← talk to server
← wait for response from server
→ talk to server
← wait for response from server
.....
Sometimes a sleep time is necessary on the client side to wait for the server to be ready loading. Except for the first example, most of these are condensed snippets from GTK-Server from www.gtk-server.org. The basic program logic will be the same for any other application.
The process fucntion allows specifying 2 pipes for communications with the launched application.
; setup communications
(map set '(myin tcout) (pipe))
(map set '(tcin myout) (pipe))
(process "/usr/bin/wish" tcin tcout)
; make GUI
(write-buffer myout
[text]
wm geometry . 250x90
wm title . "Tcl/Tk and newLISP"
bind . <Destroy> {puts {(exit)}}
[/text])
; run event loop
(while (read-line myin)
(eval-string (current-line))
)
This is the preferred way to set up longer lasting, bidirectional communications with Unix command line utilities and languages. For one-command exchanges the exec function does the job shorter.
For a more elaborate Tcl/Tk example see the application examples/tcltk.lsp in the source distribution.
; Define communication function
(define (gtk str , tmp)
(net-send connection str)
(net-receive connection tmp 64)
tmp)
; Start the gtk-server
(process "gtk-server tcp localhost:50000")
(sleep 1000)
; Connect to the GTK-server
(set 'connection (net-connect "localhost" 50000))
(set 'result (gtk "gtk_init NULL NULL"))
(set 'result (gtk "gtk_window_new 0"))
.....
Make a FIFO first (looks like a special file node):
(exec "mkfifo myfifo")
;; or alternatively
(import "/lib/libc.so.6" "mkfifo")
(mkfifo "/tmp/myfifo" 0777)
; Define communication function
(define (gtk str)
(set 'handle (open "myfifo" "write"))
(write-buffer handle str)
(close handle)
(set 'handle (open "myfifo" "read"))
(read-buffer handle tmp 20)
(close handle)
tmp)
Note that the listen function with "udp" option just binds the sockets to a address/hardware but not actually listens as in TCP/IP.
; Define communication function
(define (gtk str , tmp)
(net-send-to "localhost" 50000 str socket)
(net-receive socket 'tmp net-buffer)
tmp)
; Start the gtk-server
(define (start)
(process "gtk-server udp localhost:50000")
(sleep 500)
(set 'socket (net-listen 50001 "localhost" "udp")) )
(set 'result (gtk "gtk_init NULL NULL"))
(set 'result (gtk "gtk_window_new 0"))
.....
This is frequently used from newLISP's interactive command line to execute processes in a blocking fashion, which need a shell to run:
(! "ls -ltr")
There is an interesting variant of this form working not inside a newLISP expression, but only on the command line:
!ls -ltr
The ! should be the first character on the command line. This form works like a shell escape in the VI editor. It is useful for invoking an editor or doing quick shell work without completely leaving the newLISP console.
(exec "ls /") → ("bin" "etc" "home" "lib")
(exec "script.cgi" cgi-input)
In this example cgi-input could contain a string feeding a query input, normally coming from a web server. Note that output in this case is written directly to the screen, and cannot be returned to newLISP. Use process and pipe for two way std i/o communications with other applications.
Shared memory, semaphores and processes work frequently together. Semaphores can synchronize tasks in different process threads and shared memory can be used to communicate between them.
The following is a more complex example showing the working of all three mechanisms at the same time.
The producer loops through all n values from i = 0 to n - 1 and puts each value into shared memory where it is picked up by the consumer thread. Semaphores are used to signal that a data value is ready for reading.
#!/usr/bin/newlisp
# prodcons.lsp - Producer/consumer
#
# usage of 'fork', 'wait-pid', 'semaphore' and 'share'
(when (= ostype "Win32")
(println "this will not run on Win32")
(exit))
(constant 'wait -1 'sig 1 'release 0)
(define (consumer n)
(set 'i 0)
(while (< i n)
(semaphore cons-sem wait)
(println (set 'i (share data)) " <-")
(semaphore prod-sem sig))
(exit))
(define (producer n)
(for (i 1 n)
(semaphore prod-sem wait)
(println "-> " (share data i))
(semaphore cons-sem sig))
(exit))
(define (run n)
(set 'data (share))
(share data 0)
(set 'prod-sem (semaphore)) ; get semaphores
(set 'cons-sem (semaphore))
(set 'prod-pid (fork (producer n))) ; start processes
(set 'cons-pid (fork (consumer n)))
(semaphore prod-sem sig) ; get producer started
(wait-pid prod-pid) ; wait for processes to finish
(wait-pid cons-pid) ;
(semaphore cons-sem release) ; release semaphores
(semaphore prod-sem release))
(run 10)
(exit)
On multiprocessor CPUs the operating system will distribute processes and child processes created on to different processor cores in an optimized fashion. newLISP offers a simple API which does all the work of launching processes and does the synchronized collection of evaluation results. The Cilk API consists of only 3 function calls implemented in newLISP as spawn, sync and abort
Since v.10.1 newLISP's message function enables communications between parent and child processes. For more details about this see the next chapter Message exchange between processes.
; calculate primes in a range
(define (primes from to)
(local (plist)
(for (i from to)
(if (= 1 (length (factor i)))
(push i plist -1)))
plist))
; start child processes
(set 'start (time-of-day))
(spawn 'p1 (primes 1 1000000))
(spawn 'p2 (primes 1000001 2000000))
(spawn 'p3 (primes 2000001 3000000))
(spawn 'p4 (primes 3000001 4000000))
; wait for a maximum of 60 seconds for all tasks to finish
(sync 60000) ; returns true if all finished in time
; p1, p2, p3 and p4 now each contain a lists of primes
The example shows, how a range of prime number generation is split up for parallel processing in three sub-ranges. All spawn calls will return immediately, but sync will block until all three child processes have finished and the result lists are available in the four variables p1 to p4.
When the timeout value specified is too short for all prcesses to finish, sync will return nil. This can be use to watch progess:
; print a dot after each 2 seconds of waiting
(until (sync 2000) (println "."))
When sync is called without parameters, it returns a list of still active process ids:
; show a list of pending process ids after
;each three-tenths of a second
(until (sync 300) (println (sync)))
(define (fibo n)
(local (f1 f2)
(if(< n 2) 1
(begin
(spawn 'f1 (fibo (- n 1)))
(spawn 'f2 (fibo (- n 2)))
(sync 10000)
(+ f1 f2)))))
(fibo 7) → 21
When processes launched with spawn finish, an inlet function specified in the sync statement can be called.
(define (report pid)
(println "process: " pid " has returned"))
; call the report function, when a child returns
(sync 10000 report)
Parent and child processes started with spawn can exchange messages. Messages flow either from the parent to child processes or from child processes to the parent. By means of evaluating messsages in the parent process, the parent process can be used as a proxy routing message between peer child processes.
The message calls itself are non-blocking returning nil on failure. They can be made blocking using loops:
; blocking sender
(until (send pid msg)) ; returns true after delivery
; blocking receiver
(until (receive pid msg)) ; returns true after pickup
The parent process loops through a list of child process ids and looks for message notifications. In the following example message calls are blocking and synchronous.
; child process transmits random numbers
(define (child-process)
(set 'ppid (sys-info -4)) ; get parent pid
(while true
(until (send ppid (rand 100))))
)
; parent starts 5 child processes, listens and displays
(dotimes (i 5) (spawn 'result (child-process)))
(for (i 1 3)
(dolist (cpid (sync)) ; iterate thru child pids
(until (receive cpid msg))
(print "pid:" cpid "->" (format "%-2d " msg)))
(println)
)
(abort) ; cancel child-processes
(exit)
generates this output:
pid:53181->47 pid:53180->61 pid:53179->75 pid:53178->39 pid:53177->3 pid:53181->59 pid:53180->12 pid:53179->20 pid:53178->77 pid:53177->47 pid:53181->6 pid:53180->56 pid:53179->96 pid:53178->78 pid:53177->18
Neither the sending child process nor the receiving parent process block. Each sends and receives messages as fast as possible.
; child process transmits random numbers non-blocking
; not all calls succeed
(set 'start (time-of-day))
(define (child-process)
(set 'ppid (sys-info -4)) ; get parent pid
(while true
(until (send ppid (rand 100))))
)
; parent starts 5 child processes, listens and displays
(dotimes (i 5) (spawn 'result (child-process)))
(set 'N 1000)
(until finished
(if (= (inc counter) N) (set 'finished true))
(dolist (cpid (sync)) ; iterate thru child pids
(receive cpid msg)
(if msg (print "pid:" cpid "->" (format "%-2d \r" msg))))
)
(abort) ; cancel child-processes
(sleep 300)
(exit)
A messaging statement can be made to block for a certain time:
(define (receive-timeout pid msec) (let ( (start (time-of-day)) (msg nil)) (until (receive pid msg) (if (> (- (time-of-day) start) 1000) (throw-error "timeout"))) msg) ) ; use it (receive-timeout pid 1000) ; returns the message or throws error
In this example blocking will occcur for 1000 ms. Many methods exist to implement timeout behavior.
Messages sent can contain expressions which can be evaluated in the recipients environment. This way variables can be set, messages routed to other processes. The following example implements a message router:
; sender process of the message
(set 'pidA (spawn 'result
(begin
(dotimes (i 3)
(set 'ppid (sys-info -4))
(set 'pid (sys-info -3))
(set 'msg '(until (send pidB (string "greetings from " pidA))))
(until (send ppid msg))) ; send expression in msg
; make parent exit until loop
(until (send ppid '(set 'finished true)))
)))
; final message recipient
(set 'pidB (spawn 'result
(begin
(set 'ppid (sys-info -4))
(set 'pid (sys-info -3))
(while true
(until (receive ppid msg))
(println pid " received " msg))
)))
; receive and evaluate messages from pidA
(until finished (if (receive pidA msg) (eval msg)))
(abort) (sleep 300)
(exit)
The expression:
(until (send pidB (string "greetings from " pidA)))
does get send from pidA to the parent, where it is evaluated, causing a message to be sent to pidB. The parent process acts as a proxy agent for process with pidA.
(until (send ppid '(set 'finished true)))
The expression (set 'finished true) is sent to the parent where it gets evaluated and causes the parent's until loop to finish.
newLISP uses symbol trees to implement hashes. Symbols in name spaces can also be serialized to a file, as will be shown in the chapter about symbol creation and lookup.
The association list is a classic LISP data structure for storing information for associative retrieval:
;; creating association lists
;; pushing at the end with -1 is optimized and
;; as fast as pushing in front
(push '("John Doe" "123-5555" 1200.00) Persons -1)
(push '("Jane Doe" "456-7777" 2000.00) Persons -1)
.....
Persons → (
("John Doe" "123-5555" 1200.00)
("Jane Doe" "456-7777" 2000.00) ...)
;; access/lookup data records
(assoc "John Doe" Persons)
→ ("John Doe" "123-5555" 1200.00 male)
(assoc "Jane Doe" Persons)
→ ("Jane Doe" "456-7777" 2000.00 female)
newLISP has a lookup function similar to what is used in spreadsheet software. This function which works like a combination of assoc and nth can find the association and pick a specific member of the data record at the same time:
(lookup "John Doe" Persons 0) → "123-555"
(lookup "John Doe" Persons -1) → make
(lookup "Jane Doe" Persons 1) → 2000.00
(lookup "Jane Doe" Persons -2) → 2000.00
;; update data records
(setf (assoc "John Doe" Persons)
'("John Doe" "123-5555" 900.00 male))
;; replace as a function of existing/replaced data
(setf (assoc "John Doe" Persons) (update-person $it))
;; delete data records
(replace (assoc "John Doe" Persons) Persons)
If the data part of an association is itself an association llist, we have a nested association:
(set 'persons '(
("Anne" (address (country "USA") (city "New York")))
("Jean" (address (country "France") (city "Paris")))
))
A different syntax of the assoc function can be used to specify multiple keys:
; one key
(assoc "Anne" persons) → ("Anne" (address (country "USA") (city "New York")))
; two keys
(assoc '("Anne" 'address) persons) → (address (country "USA") (city "New York"))
; three keys
(assoc '("Anne" address city) persons) → (city "New York")
; three keys in a vector
(set 'anne-city '("Anne" address city))
(assoc anne-city persons) → (city "New York")
When all keys are symbols, as is in address, country and city, simple and nested associations in newLISP have the same format as newLISP FOOP (Functional Object Oriented Programming) objects. See the users manual chapter "17. Object Oriented Programming in newLISP" for details.
The functions assoc and setf can be used to update simple or nested associations:
(setf '("Anne" address city) persons '(city "Boston")) → (city "New York")
setf always returns the newly set element.
Hashes and FOOP objects can be combined to form an in-memory database with keyed access.
In the following example data records are stored in a hash namespace and access with the name of the person as a key.setf and lookup are used to update nested FOOP objects:
(new Tree 'Person) (new Class 'Address) (new Class 'City) (new Class 'Telefone) (Person "John Doe" (Address (City "Small Town") (Telefone 5551234))) (lookup 'Telefone (Person "John Doe")) (setf (lookup 'Telefone (Person "John Doe")) 1234567) (setf (lookup 'City (Person "John Doe")) (lower-case $it)) (Person "John Doe") → (Address (City "small town") (Telefone 1234567))
Many of todays applications are distributed on to several computers on the network or distibuted on to several processes on one CPU. Often both methods of distributing an application are used at the same time.
newLISP has facilities to evaluate many expressions in parallel on different network nodes or processes running newLISP. The net-eval function does all the work necessary to communicate to other nodes, distribute expressions for evaluation and collect results in either a blocking or event driven fashion.
The functions read-file, write-file, append-file and delete-file can also be used to access with files on remote nodes when using URLs in file specifications. In a similar way the functions load and save can be used to load and save code from and to remote nodes.
newLISP uses existing HTTP protocols and newLISP command line behavior to implement this functionality. This means that programs can be debugged and tested using standard UNIX applications like terminal, telnet or a web browser. This also enables easy integration of other tools and programs into distributed applications built with newLISP. For example the UNIX utility netcat (nc) could be used to evaluate expressions remotely or a web browser could be used to retrieve webpages from nodes running a newLISP server.
A newLISP server node is essentially a newLISP process listening to a network port and behaving like a newLISP command-line console and HTTP server for HTTP GET, PUT, POST and DELETE requests. Since version 9.1 newLISP server mode also answers CGI queries received by either GET or POST request.
Two methods are used to start a newLISP server node. One results in a state full server, maintaining state in between communications with different clients, the other method a server with no state, reloading for every new client connection.
Start a newLISP state-full server
newlisp -c -d 4711 &
newlisp myprog.lsp -c -d 4711 &
newlisp myprog.lsp -c -w /home/node25 -d 4711 &
newLISP is now listening on port 4711, the & (ampersand) sign tells newLISP to run in the background (UNIX only). The -c switch suppresses command line prompts. newLISP now behaves like a newLISP console without prompts listening on port 4711 for command line like input. Any other available port could have been chosen. Note that on UNIX ports below 1024 need administrator access rights.
The second example also pre-loads code. The third example also specifies a working directory using the -w option. If no working directory is specified using -w the startup directory is assumed to be the working directory.
Start a newLISP inetd stateless serverOn UNIX the inetd or xindetd facility can be used to start a stateless server. In this case the TCP/IP net connections are managed by a special UNIX utility with the ability to handle multiple requests at the same time. For each connection made by a client the inetd or xinetd utility will start a fresh newLISP process. After the connection is closed the newLISP process will shut down.
When nodes are not required to keep state, this is the preferred method for a newLISP server node, for handling multiple connections at the same time.
The inetd or xinetd process needs to be configured using configuration files found in the /etc directory of most UNIX installations.
For both the inetd and xinetd configurations add the following line to the /etc/services file:
net-eval 4711/tcp # newLISP net-eval requests
Note that any other port than 4711 could be supplied.
When configuring inetd add also the following lines to the /etc/inetd.conf file:
net-eval stream tcp nowait root /usr/bin/newlisp -c
# as an alternative, a program can also be preloaded
net-eval stream tcp nowait root /usr/bin/newlisp myprog.lsp -c
# a working directory can also be specified
net-eval stream tcp nowait newlisp /usr/bin/newlisp -c -w /usr/home/newlisp
The last line also specified a working directory and a user newlisp instead of the root user. This is a more secure mode limiting newLISP server node access to a specific user account with restricted permissions.
On some UNIX system a modern flavor of inetd: the xinetd facility can be used. Add the following configuration to a file /etc/xinet.d/net-eval:
service net-eval
{
socket_type = stream
wait = no
user = root
server = /usr/bin/newlisp
port = 4711
server_args = -c -w /home/node
}
Note that a variety of parameter combinations are possible to restrict access from different places or limit access to certain users. Consult the man-pages for inetd and xinetd for details.
After configuring inetd or xinetd either process must be restarted to re-read the configuration files. This can be accomplished by sending the UNIX HUP signal to either the inetd or xinetd process using the UNIX kill or UNIX nohup utility.
On Mac OS X the launchd facility can be used in a similar fashion. The newLISP source distribution contains the file org.newlisp.newlisp.plist in the util/ directory. This file can be used to launch newlisp server during OS boot time as a persisten server.
Testing the server using telnetA newLISP server node can be tested using the UNIX telnet utility:
telnet localhost 4711
; or when running on a different computer i.e. ip 192.168.1.100
telnet 192.168.1.100 4711
Multi-line expressions can be entered by enclosing them in [cmd], [/cmd] tags, each tag on a separate line. Both, the opening and closing tags should be on separate lines.
Testing the server using netcat (named nc on most UNIX)
echo '(symbols) (exit)' | nc localhost 4711
Or talking to a remote node:
echo '(symbols) (exit)' | nc 192.168.1.100 4711
In both examples netcat will echo back the result of evaluating the (symbols) expression.
Multi-line expressions can be entered by enclosing them in [cmd], [/cmd] tags, each tag on a separate line.
Testing the server from a newLISP command lineThe net-eval function as a syntax form for connecting to only one remote server node. This mode is practical for quick testing from the newLISP command line:
(net-eval "localhost" 4711 "(+ 3 4)" 1000) → 7
; to a remote node
(net-eval "192.168.1.100" 4711 {(upper-case "newlisp")} 1000) → "NEWLISP"
In the second example curly braces {,} are used to limit the program string for evaluation. This way quotes can be used to limit a string inside the expression.
No [cmd], [/cmd] tags are required when sending multi-line expressions. net-eval supplies these tags automatically.
Testing the server HTTP mode using a web browserA newLISP server also understands simple HTTP GET and PUT requests. Enter the full path of a file in the address-bar of the browser:
http://localhost:4711//usr/share/newlisp/doc/newlisp_manual.html
The manual file is almost 800 Kbyte in size and will take a few seconds to load into the browser. Specify the port-number with a colon separated from the host-name or host IP. Note the double slash // necessary to specify a file address relative to the root directory.
When testing the correct installation of newLISP server nodes, we were already sending expressions to remote node for evaluation. Many times remote evaluation is used to split up a lengthy task into shorter subtasks for remote evaluation on different nodes.
The first example is trivial, because it only evaluates several very simple expressions remotely, but it shows the principles involved in a way easy to understand:
#!/usr/bin/newlisp
(set 'result (net-eval '(
("192.168.1.100" 4711 {(+ 3 4)})
("192.168.1.101" 4711 {(+ 5 6)})
("192.168.1.102" 4711 {(+ 7 8)})
("192.168.1.103" 4711 {(+ 9 10)})
("192.168.1.104" 4711 {(+ 11 12)})
) 1000))
(println "result: " result)
(exit)
Running this program will produce the following output:
result: (7 11 15 19 23)
When running UNIX and using an inetd or xinetd configured newLISP server, the servers and programs can be run on just one CPU replacing all IP numbers with "localhost" or the same local IP number. The indetd or xinetd daemon will then start 5 independent newLISP processes. On Win32 5 state-full newLISP servers could be started on different port numbers to accomplish the same.
Instead of collecting all results at one on the return of net-eval, a call back function can be used to receive and process results as they become available:
#!/usr/bin/newlisp
(define (idle-loop p)
(if p (println p)))
(set 'result (net-eval '(
("192.168.1.100" 4711 {(+ 3 4)})
("192.168.1.101" 4711 {(+ 5 6)})
("192.168.1.102" 4711 {(+ 7 8)})
("192.168.1.103" 4711 {(+ 9 10)})
("192.168.1.104" 4711 {(+ 11 12)})
) 1000 idle-loop))
(exit)
While net-eval is waiting for results it calls the function idle-loop repeatedly with parameter p. The parameter p is nil when no result was received during the last 100 micro seconds or p contains a list sent back from the remote node. The list contains the remote address and port and the evaluation result. The example shown would generate the following output:
("192.168.1.100" 4711 7)
("192.168.1.101" 4711 11)
("192.168.1.102" 4711 15)
("192.168.1.103" 4711 19)
("192.168.1.104" 4711 23)
For testing on just one CPU replace addresses with "localhost" the UNIX inetd or xinetd daemon will start a separate process for each connection made and all listening on port 4711. When using a state-full server on the same Win32 CPU specify a different port number for each server
Setting up the net-eval parameter structureIn a networked environment where an application gets moved around, or server nodes with changing IP numbers are used, it is necessary to set up the node parameters in the net-eval parameter list as variables. The following more complex example shows how this can be done. The example also shows how a bigger piece of program text can be transferred to a remote node for evaluation and how this program piece can be customized for each node differently:
#!/usr/bin/newlisp
; node parameters
(set 'nodes '(
("192.168.1.100" 4711)
("192.168.1.101" 4711)
("192.168.1.102" 4711)
("192.168.1.103" 4711)
("192.168.1.104" 4711)
))
; program template for nodes
(set 'program [text]
(begin
(map set '(from to node) '(%d %d %d))
(for (x from to)
(if (= 1 (length (factor x)))
(push x primes -1)))
primes)
[/text])
; call back routine for net-eval
(define (idle-loop p)
(if p
(begin
(println (p 0) ":" (p 1))
(push (p 2) primes))))
(println "Sending request to nodes, and waiting ...")
; machines could be on different IP addresses.
; For this test 5 nodes are started on localhost
(set 'result (net-eval '(
((nodes 0 0) (nodes 0 1) (format program 0 99999 1))
((nodes 1 0) (nodes 1 1) (format program 100000 199999 2))
((nodes 2 0) (nodes 2 1) (format program 200000 299999 3))
((nodes 3 0) (nodes 3 1) (format program 300000 399999 4))
((nodes 4 0) (nodes 4 1) (format program 400000 499999 5))
) 20000 idle-loop))
(set 'primes (sort (flat primes)))
(save "primes" 'primes)
(exit)
At the beginning of the program a nodes list structure contains all the relevant node information for hostname and port.
The program calculates all prime numbers in a given range. The from, to and node variables are configured into the program text using format. All instructions are placed into a begin expression block, so only one expression result will be send back from the remote node.
Many other schemes to configure a net-eval parameter list are possible. The example shows, that net-eval evaluates the node parameter specifications inside the quoted list. The following scheme would give the same resuls:
(set 'node-eval-list '(
((nodes 0 0) (nodes 0 1) (format program 0 99999 1))
((nodes 1 0) (nodes 1 1) (format program 100000 199999 2))
((nodes 2 0) (nodes 2 1) (format program 200000 299999 3))
((nodes 3 0) (nodes 3 1) (format program 300000 399999 4))
((nodes 4 0) (nodes 4 1) (format program 400000 499999 5))
))
(set 'result (net-eval node-eval-list 20000 idle-loop))
The function idle-loop aggregates all lists of primes received and generates the following output:
192.168.1.100:4711
192.168.1.101:4711
192.168.1.102:4711
192.168.1.103:4711
192.168.1.104:4711
As with the previous examples all IP numbers could be replaced with "localhost" or any other host-name or IP number to test a distributed application on a single host before deployment in a distributed environment with many networked hosts.
Files can be read from or written to remote nodes with the same functions used to read and write files to a local file system. This functionality is currently only available on UNIX systems when talking to newLISP servers. As functions are based on standard GET and PUT HTTP protocols they can also be used communicating with web servers. Note that few Apache web-server installations have enabled the PUT protocol by default.
The functions read-file, write-file and append-file can all take URLs in their filename specifications for reading from and writing to remote nodes running a newLISP server or a web-server:
(write-file "http://127.0.0.1:4711//Users/newlisp/afile.txt" "The message - ")
→ "14 bytes transferred for /Users/lutz/afile.txt\r\n"
(append-file "http://127.0.0.1:4711//Users/newlisp/afile.txt" "more text")
→ "9 bytes transferred for /Users/lutz/afile.txt\r\n"
(read-file "http://127.0.0.1:4711//Users/newlisp/afile.txt")
→ "The message - more text"
The first two function return a message starting with the numbers of bytes transferred and the name of the remote file affected. The -read-file function returns the contents received.
Under all error conditions an error message starting with the characters ERR: would be returned:
(read-file "http://127.0.0.1:4711//Users/newlisp/somefile.txt")
→ "ERR:404 File not found: /Users/newlisp/somefile.txt\r\n"
Note the double backslash necessary to reference files relative to root on the server node.
All functions can be used to transfer binary non-ascii contents containing zero characters. Internally newLISP uses the functions get-url and put-url, which could be used instead of the functions read-file, write-file and append-file. Additional options like used with get-url and put-url could be used with the functions read-file, write-file and append-file as well. For more detail see the newLISP function reference for these functions.
The same load and save functions used to load program or LISP data from a local file system can be used to load or save programs and LISP data from or to remote nodes.
By using URLs in the file specifications of load and save these functions can work over the network communicating with a newLISP server node.:
(load "http://192.168.1.2:4711//usr/share/newlisp/mysql5.lsp")
(save "http://192.168.1.2:4711//home/newlisp/data.lsp" 'db-data)
Although the load and save functions internally use get-url and put-url to perform its works they behave exactly as when used on a local file system, but instead of a file path URLs are specified. Both function will timeout after 60 seconds if a connection could not be established. When finer control is necessary use the functions get-url and put-url together with eval-string and source to realize a similar result as when using the load and save in HTTP mode.
newLISP supports named local domain sockets in newLISP server mode and using the built-in functions net-eval, net-listen, net-connect together with the functions net-accept, net-receive, net-select and net-send.
Using local domain sockets fast communications between processes on the same file system and with newLISP servers is possible. See the Users Manual for more details.
In all previous chapters the -c server mode was used. This mode can act as a net-eval server and at the same time answer HTTP requests for serving web pages or transfer of files and programs. The -c mode is the preferred mode for secure operation behind a firewall. newLISP also has a -http mode which works like a restricted -c mode. In -http mode only HTTP requests are served and command-line like formatted requests and net-eval requests are not answered. In this mode newLISP can act like a web server answering HTTP GET, PUT, POST and DELETE requests as well as CGI requests, but additional efforts should be made to restrict the access to unauthorized files and directories to secure the server when exposed to the internet.
When newLISP server answers any kind of requests (HTTP and command line), the newLISP function command-event can be used to pre-process the request. The pre-processing function can be loaded from a file httpd-conf.lsp when starting the sever:
server_args = httpd-conf.lsp -http -w /home/node
The above snippet shows part of a xinetd configuration file. A startup program httpd-conf.lsp has been added which will be loaded upon invocation of newLISP. The -c options has been replaced with the -http option. Now neither net-eval nor command-line requests will be answered but only HTTP requests.
The startup file could also have been added the following way when starting the server in the background from a command shell, and httpd-conf.lsp is in the current directory:
newlisp httpd-conf.lsp -http -d 80 -w /home/www &
All requests will be pre-processed with a function specified using command-event in httpd-conf.lsp:
;; httpd-conf.lsp
;;
;; filter and translate HTTP request for newLISP
;; -c or -http server modes
;; reject query commands using CGI with .exe files
(command-event (fn (s)
(local (request)
(if (find "?" s) ; is this a query
(begin
(set 'request (first (parse s "?")))
; discover illegal extension in queries
(if (ends-with request ".exe")
(set 'request "GET /errorpage.html")
(set 'request s)))
(set 'request s))
request)
))
; eof
All CGI requests files ending with .exe would be rejected and the request translated into the request of an error page.
On http://www.newlisp.org various CGI examples can be found. In the download directory at http://www.newlisp.org/downloads two more complex applications can be found: newlisp-ide is a web based IDE and newlisp-wiki is a content management system which also runs the www.newlisp.org website.
The following is a minimum CGI program:
#!/usr/bin/newlisp
(print "Content-type: text/html\r\n\r\n")
(println "<h2>Hello World</h2>")
(exit)
newLISP normally puts out a standard HTTP/1.0 200 OK\r\n response header plus a Server: newLISP v. ...\r\n header line. If the first line of CGI program output starts with "Status:" then newLISP's standard header output is suppressed, and the CGI program must supply the full header by itself. The following example redirects a request to a new location:
#!/usr/bin/newlisp
(print "Status: 301 Moved Permanently\r\n")
(print "Location: http://www.newlisp.org/index.cgi\r\n\r\n")
(exit)
A newLISP installation contains a module file cgi.lsp. This module contains subroutines for extracting parameters from HTTP GET and POST requests, extract or set cookies amd other useful routines when writing CGI files. See the modules section at: http://www.newlisp.org/modules/.
In both the -c and -http HTTP modes the following file types are recognized and a correctly formatted Content-Type: header is sent back:
file extension media type .html text/html .htm text/htm .css text/css .gif image/gif .png image/png .mp3 audio/mpeg .wav audio/x-wav .mpg video/mpeg .mov video/quicktime .avi video/x-msvideo application/pdf .js application/javascript .zip application/zip any other text/plain media types in newLISP HTTP request handling
When receiving HTTP requests newLISP server mode will extract information from the HTTP request header and configure HTTP_HOST, HTTP_USER_AGENT and HTTP_COOKIE in the environment of the host operating system. If a httpd-conf functions was loaded on newLISP server startup header processing will happen after processing httpd-conf. The environment variables can be accessed by a CGI process.
newLISP has an import function, which allows importing function from DLLs (Dynamic Link Libraries) on Win32 or shared libraries on Linux/UNIX (ending in .so).
This chapter shows how to compile and use libraries on both, Win32 and Linux/UNIX platforms. We will compile a DLL and a Linux/UNIX shared library from the following 'C' program:
#include <stdio.h>
#include <stdlib.h>
#include <ctype.h>
int foo1(char * ptr, int number)
{
printf("the string: %s the number: %d\n", ptr, number);
return(10 * number);
}
char * foo2(char * ptr, int number)
{
char * upper;
printf("the string: %s the number: %d\n", ptr, number);
upper = ptr;
while(*ptr) { *ptr = toupper(*ptr); ptr++; }
return(upper);
}
/* eof */
Both functions foo1 and foo2 print their arguments, but while foo1 returns the number multiplied 10 times, foo2 returns the uppercase of a string to show how to return strings from 'C' functions.
On Linux/UNIX we can compile and link testlib.so in one step:
gcc testlib.c -shared -o testlib.so
Or On Mac OSX/darwin do:
gcc -bundle -o testlib.so
The library testlib.so will be built with Linux/UNIX default cdecl conventions. Importing the library is very similar on both Linux and Win32 platforms, but on Win32 the library can be found in the current directory. You may have to specify the full path or put the library in the library path of the os:
newLISP v.8.4.3 on Linux, execute 'newlisp -h' for more info.
> (import "/home/newlisp/testlib.so" "foo1")
foo1 <6710118F>
> (import "/home/newlisp/testlib.so" "foo2")
foo2 <671011B9>
> (foo1 "hello" 123)
the string: hello the number: 123
1230
> (foo2 "hello" 123)
the string: hello the number: 123
4054088
> (get-string (foo2 "hello" 123))
the string: hello the number: 123
"HELLO"
>
Again, the number returned from foo2 is the string address pointer and get-string can be used to access the string. When using get-string only character up to a zero byte are returned. When returning the addresses to binary buffers different techniques using unpack are used to access the information.
DLLs on Win32 can be made using the MinGW, Borland or CYGWIN compilers. This example shows, how to do it using the MinGW compiler.
Compile it:
gcc -c testlib.c -o testlib.o
Before we can transform testlib.o into a DLL we need a testlib.def declaring the exported functions:
LIBRARY testlib.dll
EXPORTS
foo1 @1 foo1
foo2 @2 foo2
Now wrap the DLL:
dllwrap testlib.o --def testlib.def -o testlib.dll -lws2_32
The library testlib.dll will be built with default Win32 stdcall conventions. The following shows an interactive session, importing the library and using the functions:
newLISP v.8.4.3 on Win32 MinGW, execute 'newlisp -h' for more info.
> (import "testlib.dll" "foo1")
foo1 <6710118F>
> (import "testlib.dll" "foo2")
foo2 <671011B9>
> (foo1 "hello" 123)
the string: hello the number: 123
1230
> (foo2 "hello" 123)
the string: hello the number: 123
4054088
> (get-string (foo2 "hello" 123))
the string: hello the number: 123
"HELLO"
>
; import a library compiled for cdecl
; calling conventsions
> (import "foo.dll" "func" "cdecl")
Note that the first time using foo2 the return value 4054088 is the memory address of the string returned. Using get-string the string belonging to it can be accessed. If the library is compiled using cdecl calling conventions, the cdecl keyword must be used in the import expression.
Just like 'C' strings are returned using string pointers, 'C' structures can be returned using structure pointers and functions like get-string, get-int or get-char can be used to access the members. The following example illustrates this:
typedef struct mystruc
{
int number;
char * ptr;
} MYSTRUC;
MYSTRUC * foo3(char * ptr, int num )
{
MYSTRUC * astruc;
astruc = malloc(sizeof(MYSTRUC));
astruc->ptr = malloc(strlen(ptr) + 1);
strcpy(astruc->ptr, ptr);
astruc->number = num;
return(astruc);
}
The newLISP program would access the structure members as follows:
> (set 'astruc (foo3 "hello world" 123))
4054280
> (get-string (get-integer (+ astruc 4)))
"hello world"
> (get-integer astruc)
123
>
The return value from foo is the address to the structure astruc. To access the string pointer, 4 must be added as the size of an integer type in the 'C' programming language. The string in the string pointer then gets accessed using get-string.
Sometimes data structures contain data types of different length than the normal CPU register word:
sruct mystruct
{
short int x;
int z;
short int y;
} data;
struct mystruct * foo(void)
{
data.x = 123;
data.y = 456;
data.z = sizeof(data);
return(&data);
}
The x and y variables are 16 bit wide and only z takes 32 bit. When a compiler on a 32-bit CPU packs this structure the variables x and y will each fill up 32 bits instead of the 16 bit each. This is necessary so the 32-bit variable z can be aligned properly. The following code would be necessary to access the structure members:
> (import "/usr/home/nuevatec/test.so" "foo")
foo <281A1588>
> (unpack "lu lu lu" (foo))
(123 12 456)
The whole structure consumes 3 by 4 = 12 bytes, because all members have to be aligned to 32 bit borders in memory.
The following data structure packs the short 16 bit variables next to each other. This time only 8 bytes are required: 2 each for x and y and 4 bytes for z. Because x and y are together in one 32-bit word, none of the variables needs to cross a 32-bit boundary:
struct mystruct
{
short int x;
short int y;
int z;
} data;
struct mystruct * foo(void)
{
data.x = 123;
data.y = 456;
data.z = sizeof(data);
return(&data);
}
This time the access code in newLISP reflects the size of the structure members:
> (import "/usr/home/nuevatec/test.so" "foo")
foo <281A1588>
> (unpack "u u lu" (foo))
(123 456 8)
| Data Type | newLISP call | C Function call |
|---|---|---|
| integer | (foo 123) | foo(int number) |
| double float | (foo 1.234) | foo(double number) |
| float | (foo (flt 1.234)) | foo(float number) |
| string | (foo "Hello World!") | foo(char * string) |
| integer array | (foo (pack "d d d" 123 456 789)) | foo(int numbers[]) |
| float array | (foo (pack "f f f" 1.23 4.56 7.89)) | foo(float[]) |
| double array | (foo (pack "lf lf lf) 1.23 4.56 7.89) | foo(double[]) |
| string array | (foo (pack "lu lu lu" "one" "two" "three"))) | foo(char * string[]) |
Note that floats and double floats are only passed correctly on x86 platforms with cdecl calling conventions or when passed by pointer reference as in variable argument functions, i.e: printf().
pack can receive multiple arguments after the format specifier in a list too:
(pack "lu lu lu" '("one" "two" "three"))
| Data Type | newLISP to extract return value | C return |
|---|---|---|
| integer | (set 'number (foo x y z)) | return(int number) |
| double float | n/a - only 32bit returns, use double float pointer instead | not available |
| double float ptr | (set 'number (get-float (foo x y z))) | return(double * numPtr) |
| float | not available | not available |
| string | (set 'string (get-string (foo x y z) | return(char * string) |
| integer array | (set 'numList (unpack "ld ld ld" (foo x y z))) | return(int numList[]) |
| float array | (set 'numList (unpack "f f f" (foo x y z))) | return(float numList[]) |
| double array | (set 'numList (unpack "lf lf lf") (foo x y z))) | return(double numList[]) |
| string array | (set 'stringList (map get-string (unpack "ld ld ld" (foo x y z)))) | return(char * string[]) |
Floats and doubles can only be returned via address pointer references.
When returning array types the number of elements in the array must be known. The examples always assume 3 elements.
All pack and unpack and formats can also be given without spaces, but are spaced in the examples for better readability.
The formats "ld" and "lu" are interchangeable, but the 16-bit formats "u" and "d" may produce different results, because of sign expansion from 16 to 32 bits.
Flags are available for changing endian byte order during pack and unpack.
Sometimes newLISP's built-in import facility cannot be used with a library. This happens whenever a library does not strictly adhere to cdecl calling conventions expecting all parameters passed on the stack. E.g. when running Mac OS X on older PPC CPUs instead of Intel CPUs, the OpenGL libararies installed by default on Mac OS X cannot be used.
A special wrapper library can be built to translate cdecl conventions expected by newLISP into the calling conventions expected by the target library.
/* wrapper.c - demo for wrapping library function */
compile:
gcc -m32 -shared wrapper.c -o wrapper.so
or:
gcc -m32 -bundle wrapper.c -o wrapper.dylib
usage from newLISP:
(import "./wrapper.dylib" "wrapperFoo")
(define (glFoo x y z)
(get-float (wrapperFoo 5 (float x) (int y) (float z))) )
(glFoo 1.2 3 1.4) => 7.8
*/
#include <stdio.h>
#include <stdarg.h>
/* the glFoo() function would normally live in the library to be
wrapped, e.g. libopengl.so or libopengl.dylib, and this
program would link to it.
For demo and test purpose the function has been included in this
file
*/
double glFoo(double x, int y, double z)
{
double result;
result = (x + z) * y;
return(result);
}
/* this is the wrapper for glFoo which gets imported by newLISP
declaring it as a va-arg function guarantees 'cdecl'
calling conventions on most architectures
*/
double * wrapperFoo(int argc, ...)
{
va_list ap;
double x, z;
static double result;
int y;
va_start(ap, argc);
x = va_arg(ap, double);
y = va_arg(ap, int);
z = va_arg(ap, double);
va_end(ap);
result = glFoo(x, y, z);
return(&result);
}
/* eof */
Version 1.2, November 2002
Copyright (C) 2000,2001,2002 Free Software Foundation, Inc. 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed.
The purpose of this License is to make a manual, textbook, or other functional and useful document "free" in the sense of freedom: to assure everyone the effective freedom to copy and redistribute it, with or without modifying it, either commercially or noncommercially. Secondarily, this License preserves for the author and publisher a way to get credit for their work, while not being considered responsible for modifications made by others.
This License is a kind of "copyleft", which means that derivative works of the document must themselves be free in the same sense. It complements the GNU General Public License, which is a copyleft license designed for free software.
We have designed this License in order to use it for manuals for free software, because free software needs free documentation: a free program should come with manuals providing the same freedoms that the software does. But this License is not limited to software manuals; it can be used for any textual work, regardless of subject matter or whether it is published as a printed book. We recommend this License principally for works whose purpose is instruction or reference.
1. APPLICABILITY AND DEFINITIONS
This License applies to any manual or other work, in any medium, that contains a notice placed by the copyright holder saying it can be distributed under the terms of this License. Such a notice grants a world-wide, royalty-free license, unlimited in duration, to use that work under the conditions stated herein. The "Document", below, refers to any such manual or work. Any member of the public is a licensee, and is addressed as "you". You accept the license if you copy, modify or distribute the work in a way requiring permission under copyright law.
A "Modified Version" of the Document means any work containing the Document or a portion of it, either copied verbatim, or with modifications and/or translated into another language.
A "Secondary Section" is a named appendix or a front-matter section of the Document that deals exclusively with the relationship of the publishers or authors of the Document to the Document's overall subject (or to related matters) and contains nothing that could fall directly within that overall subject. (Thus, if the Document is in part a textbook of mathematics, a Secondary Section may not explain any mathematics.) The relationship could be a matter of historical connection with the subject or with related matters, or of legal, commercial, philosophical, ethical or political position regarding them.
The "Invariant Sections" are certain Secondary Sections whose titles are designated, as being those of Invariant Sections, in the notice that says that the Document is released under this License. If a section does not fit the above definition of Secondary then it is not allowed to be designated as Invariant. The Document may contain zero Invariant Sections. If the Document does not identify any Invariant Sections then there are none.
The "Cover Texts" are certain short passages of text that are listed, as Front-Cover Texts or Back-Cover Texts, in the notice that says that the Document is released under this License. A Front-Cover Text may be at most 5 words, and a Back-Cover Text may be at most 25 words.
A "Transparent" copy of the Document means a machine-readable copy, represented in a format whose specification is available to the general public, that is suitable for revising the document straightforwardly with generic text editors or (for images composed of pixels) generic paint programs or (for drawings) some widely available drawing editor, and that is suitable for input to text formatters or for automatic translation to a variety of formats suitable for input to text formatters. A copy made in an otherwise Transparent file format whose markup, or absence of markup, has been arranged to thwart or discourage subsequent modification by readers is not Transparent. An image format is not Transparent if used for any substantial amount of text. A copy that is not "Transparent" is called "Opaque".
Examples of suitable formats for Transparent copies include plain ASCII without markup, Texinfo input format, LaTeX input format, SGML or XML using a publicly available DTD, and standard-conforming simple HTML, PostScript or PDF designed for human modification. Examples of transparent image formats include PNG, XCF and JPG. Opaque formats include proprietary formats that can be read and edited only by proprietary word processors, SGML or XML for which the DTD and/or processing tools are not generally available, and the machine-generated HTML, PostScript or PDF produced by some word processors for output purposes only.
The "Title Page" means, for a printed book, the title page itself, plus such following pages as are needed to hold, legibly, the material this License requires to appear in the title page. For works in formats which do not have any title page as such, "Title Page" means the text near the most prominent appearance of the work's title, preceding the beginning of the body of the text.
A section "Entitled XYZ" means a named subunit of the Document whose title either is precisely XYZ or contains XYZ in parentheses following text that translates XYZ in another language. (Here XYZ stands for a specific section name mentioned below, such as "Acknowledgements", "Dedications", "Endorsements", or "History".) To "Preserve the Title" of such a section when you modify the Document means that it remains a section "Entitled XYZ" according to this definition.
The Document may include Warranty Disclaimers next to the notice which states that this License applies to the Document. These Warranty Disclaimers are considered to be included by reference in this License, but only as regards disclaiming warranties: any other implication that these Warranty Disclaimers may have is void and has no effect on the meaning of this License.
2. VERBATIM COPYING
You may copy and distribute the Document in any medium, either commercially or noncommercially, provided that this License, the copyright notices, and the license notice saying this License applies to the Document are reproduced in all copies, and that you add no other conditions whatsoever to those of this License. You may not use technical measures to obstruct or control the reading or further copying of the copies you make or distribute. However, you may accept compensation in exchange for copies. If you distribute a large enough number of copies you must also follow the conditions in section 3.
You may also lend copies, under the same conditions stated above, and you may publicly display copies.
3. COPYING IN QUANTITY
If you publish printed copies (or copies in media that commonly have printed covers) of the Document, numbering more than 100, and the Document's license notice requires Cover Texts, you must enclose the copies in covers that carry, clearly and legibly, all these Cover Texts: Front-Cover Texts on the front cover, and Back-Cover Texts on the back cover. Both covers must also clearly and legibly identify you as the publisher of these copies. The front cover must present the full title with all words of the title equally prominent and visible. You may add other material on the covers in addition. Copying with changes limited to the covers, as long as they preserve the title of the Document and satisfy these conditions, can be treated as verbatim copying in other respects.
If the required texts for either cover are too voluminous to fit legibly, you should put the first ones listed (as many as fit reasonably) on the actual cover, and continue the rest onto adjacent pages.
If you publish or distribute Opaque copies of the Document numbering more than 100, you must either include a machine-readable Transparent copy along with each Opaque copy, or state in or with each Opaque copy a computer-network location from which the general network-using public has access to download using public-standard network protocols a complete Transparent copy of the Document, free of added material. If you use the latter option, you must take reasonably prudent steps, when you begin distribution of Opaque copies in quantity, to ensure that this Transparent copy will remain thus accessible at the stated location until at least one year after the last time you distribute an Opaque copy (directly or through your agents or retailers) of that edition to the public.
It is requested, but not required, that you contact the authors of the Document well before redistributing any large number of copies, to give them a chance to provide you with an updated version of the Document.
4. MODIFICATIONS
You may copy and distribute a Modified Version of the Document under the conditions of sections 2 and 3 above, provided that you release the Modified Version under precisely this License, with the Modified Version filling the role of the Document, thus licensing distribution and modification of the Modified Version to whoever possesses a copy of it. In addition, you must do these things in the Modified Version:
If the Modified Version includes new front-matter sections or appendices that qualify as Secondary Sections and contain no material copied from the Document, you may at your option designate some or all of these sections as invariant. To do this, add their titles to the list of Invariant Sections in the Modified Version's license notice. These titles must be distinct from any other section titles.
You may add a section Entitled "Endorsements", provided it contains nothing but endorsements of your Modified Version by various parties--for example, statements of peer review or that the text has been approved by an organization as the authoritative definition of a standard.
You may add a passage of up to five words as a Front-Cover Text, and a passage of up to 25 words as a Back-Cover Text, to the end of the list of Cover Texts in the Modified Version. Only one passage of Front-Cover Text and one of Back-Cover Text may be added by (or through arrangements made by) any one entity. If the Document already includes a cover text for the same cover, previously added by you or by arrangement made by the same entity you are acting on behalf of, you may not add another; but you may replace the old one, on explicit permission from the previous publisher that added the old one.
The author(s) and publisher(s) of the Document do not by this License give permission to use their names for publicity for or to assert or imply endorsement of any Modified Version.
5. COMBINING DOCUMENTS
You may combine the Document with other documents released under this License, under the terms defined in section 4 above for modified versions, provided that you include in the combination all of the Invariant Sections of all of the original documents, unmodified, and list them all as Invariant Sections of your combined work in its license notice, and that you preserve all their Warranty Disclaimers.
The combined work need only contain one copy of this License, and multiple identical Invariant Sections may be replaced with a single copy. If there are multiple Invariant Sections with the same name but different contents, make the title of each such section unique by adding at the end of it, in parentheses, the name of the original author or publisher of that section if known, or else a unique number. Make the same adjustment to the section titles in the list of Invariant Sections in the license notice of the combined work.
In the combination, you must combine any sections Entitled "History" in the various original documents, forming one section Entitled "History"; likewise combine any sections Entitled "Acknowledgements", and any sections Entitled "Dedications". You must delete all sections Entitled "Endorsements."
6. COLLECTIONS OF DOCUMENTS
You may make a collection consisting of the Document and other documents released under this License, and replace the individual copies of this License in the various documents with a single copy that is included in the collection, provided that you follow the rules of this License for verbatim copying of each of the documents in all other respects.
You may extract a single document from such a collection, and distribute it individually under this License, provided you insert a copy of this License into the extracted document, and follow this License in all other respects regarding verbatim copying of that document.
7. AGGREGATION WITH INDEPENDENT WORKS
A compilation of the Document or its derivatives with other separate and independent documents or works, in or on a volume of a storage or distribution medium, is called an "aggregate" if the copyright resulting from the compilation is not used to limit the legal rights of the compilation's users beyond what the individual works permit. When the Document is included in an aggregate, this License does not apply to the other works in the aggregate which are not themselves derivative works of the Document.
If the Cover Text requirement of section 3 is applicable to these copies of the Document, then if the Document is less than one half of the entire aggregate, the Document's Cover Texts may be placed on covers that bracket the Document within the aggregate, or the electronic equivalent of covers if the Document is in electronic form. Otherwise they must appear on printed covers that bracket the whole aggregate.
8. TRANSLATION
Translation is considered a kind of modification, so you may distribute translations of the Document under the terms of section 4. Replacing Invariant Sections with translations requires special permission from their copyright holders, but you may include translations of some or all Invariant Sections in addition to the original versions of these Invariant Sections. You may include a translation of this License, and all the license notices in the Document, and any Warranty Disclaimers, provided that you also include the original English version of this License and the original versions of those notices and disclaimers. In case of a disagreement between the translation and the original version of this License or a notice or disclaimer, the original version will prevail.
If a section in the Document is Entitled "Acknowledgements", "Dedications", or "History", the requirement (section 4) to Preserve its Title (section 1) will typically require changing the actual title.
9. TERMINATION
You may not copy, modify, sublicense, or distribute the Document except as expressly provided for under this License. Any other attempt to copy, modify, sublicense or distribute the Document is void, and will automatically terminate your rights under this License. However, parties who have received copies, or rights, from you under this License will not have their licenses terminated so long as such parties remain in full compliance.
10. FUTURE REVISIONS OF THIS LICENSE
The Free Software Foundation may publish new, revised versions of the GNU Free Documentation License from time to time. Such new versions will be similar in spirit to the present version, but may differ in detail to address new problems or concerns. See http://www.gnu.org/copyleft/.
Each version of the License is given a distinguishing version number. If the Document specifies that a particular numbered version of this License "or any later version" applies to it, you have the option of following the terms and conditions either of that specified version or of any later version that has been published (not as a draft) by the Free Software Foundation. If the Document does not specify a version number of this License, you may choose any version ever published (not as a draft) by the Free Software Foundation.