Luban is a component oriented programming language. Luban is named to honor a legendary ancient Chinese civil engineer two thousand year ago whose constructions are still in use today.

Luban, in short speaking, is a scripting language with a robust component and composition model. Luban is easy to use. To start coding Luban requires almost zero understanding of any high level programming concepts. If you know what you want to do plus a minimum set of Luban’s basic features, you can code your work with ease. Luban’s syntax is simple and clean. And Luban code requires no explicit compiling and linking steps to run. If you like the usability of scripting language, you will like Luban as well.

Intuitiveness and usability are the design principles of Luban. And that is why Luban has the unique component composition feature. Component composition is a technology that has been used in engineering. Yet Luban is the first programming language that directly supports component composition in software construction. Luban component composition is analogous to spreadsheet construction. User defines a new component by laying down sub-components and specifying dependencies among the sub-components. Composition fits a range of problems that could be awkward to code in traditional procedural way. More importantly, composition is simple and intuitive. Any one who can use a spreadsheet can immediately understand Luban component composition.

Luban also fully supports component oriented programming paradigm in which users can easily share and reuse component. Luban’s component design is similar to Java Bean, yet much more robust. You may ask that why not use object oriented paradigm. The answer is object oriented design is probably too complicated for general scripting language users. Scripting language needs its own component model to go further than coding ad hoc short script, and its component model should be simpler than class.

Luban’s component features include component interface definition and inheritance, component name space, built-in multi-threading, dynamic component loading and dynamic type checking. You can start using Luban only for scripting and later may find these component features handy when you try to collect your Luban scripts into one or more coherent systems.

In conclusion, Luban is an integration layer programming language that has both the usability of scripting language and the scalability of structured language. The goal of Luban is to present a simple component oriented programming tool to programmers and a potentially much wider audience that may never have been able to program before. The design is to empower people who have the field knowledge to directly code and share their knowledge in the form of software components, no matter if they are professional programmer or not.

2 Start Scripting Luban

In the beginner's mind there are many possibilities,

in the expert's mind there are few.
- Shunryu Suzuki, Zen Master

This chapter introduces the basic procedural code constructions in Luban through examples. This chapter is all you need to read for basic Luban scripting. The syntax of Luban procedural code is very much like simplified C++ or Java. Basically a valid Luban script is a series of one or more Luban statements, separated by either semicolon or bracket. Change of line doesn’t mean much for Luban. So in theory you can put all your code in one big line and claim you finish a system in a single line of code.

2.1 Hello World

Nothing better illustrates a programming language than code examples. And all self-respecting languages start with a “Hello, world” example. Luban is no exception.

std::println(obj=“Hello, world!”);

Type above line into a file named “helloworld.lbn”. Then you can run your first Luban program by type the following in your command prompt of your operating system:

Mycomputer > luban helloworld.lbn

Hello, world!

This simple “Hello, world” Luban example code does one thing that is to call one Luban component structure of type std::println with its input property “obj” set to string “Hello, world!”. This component call trigger the evaluation of std::println that prints out the “Hello, world!” message on console screen..

2.2 Procedural Statements

The basic procedural constructions of Luban consist of expressions and statements. If you know C++ or Java, you can skip most of this section except the foreach statement and multi-threading part.

2.2.1 Expression and Assignment

The most basic Luban statement can be an expression or an assignment statement. Listed below are different kinds of Luban expressions, in the order of precedence.

Constants:

1; 3.14; true; false; “Hello”; ‘a’;

Variables are also expressions. Luban’s variable name rule is the same as C/C++/Java

varx; _varx_; v123;

Data object and component construction:

[ 1,2,3, x, y, z, a+b ]; // vector construction

{ “one”: 1, “two” : 2 }; // map construction

{ “one”, “two” }; // set construction

double(“3.1415”); // double type construction

std::file(“mydatafile”, ‘r’); // open file

mynamespace::mycomponentX(); //Luban component construction

mynamespace::mycomponetX( prp1 = 100 ); // component call

Container element access, data object member function call, component property access and component object call.

mymap[“one”]; // map indexing

x[1] = y[2]; // vector or map indexing

mymap.remove(“one”); // data object member function call

linevector = allines.split(‘\n’); // data object member function call

mycalculator( input1=1.1, input2=2.2); // component object call

Single operand operators, ++ -- and negation:

++i; --i; i++; i--; // i will be changed

y = -x; // x is not changed, y get the –x result;

Common arithmetic operations:

1+2; 10.0*3.0; 50%3; 50/3; “Hello, “+”world”;

Type checking operations:

objx isa string;

typeof(objx) == string;

Equality and order checking:

x == y; x != y; x > y; x < y; x >= y; x <= y;

Logical expressions:

true or false; x and y; not true; not a and not b;

true || false; x && y; ! true; ! a && !b; // for people who know C/C++/Java

Conditional expression:

x > y ? x: y;

Assignment statements:

x = 1; x += 1; x[index] = 2; x.property = 3; x*=2; x/=3;

The details of some of the expressions like property reference and component call can be found in later chapters.

2.2.2 Flow Control

Luban has a set of common flow control statements. The below is a brief list and examples of usage.

IF statement:

if ( a>b ) result=a;

IF-ELSE statement:

if ( a>b ) result = a; else result = b;

WHILE statement:

while ( x < 100 ) ++ x;

FOR statement:

for( i=0; i<100; ++i) std::console().writeline(vec[i]);

for(;;); // forever loop

FOREACH statement:

foreach( key, value in {“one”:1, “two”:2} )

std::console().writeline(key, ‘ ‘, value);

foreach( element in [ 1,2,3 ] )

std::console().writeline(element);

BREAK statement, used to jump out of a loop:

for( i=0; ; i++) if ( i>100 ) break;

CONTINUE statement, used to jump to the beginning of a loop:

for( i=0; ; i++) if ( i<=100 ) continue; else break;

DONOTHING statement:

;

FINISH statement, terminate the code execution in current component.

finish;

WAIT statement, wait for all dispatched threads to finish.

wait;

WAITFOR statement, wait for thread(s) associated with specific variable(s) to finish.

waitfor x,y,z;

CANCEL statement, cancel and join all dispatched threads

cancel;

Details for these thread related statements can be found later in this chapter.

2.2.3 Compound Statement:

Using bracket {} to collect one or more statement together makes a compound statement. You can use compound statement anywhere you use a regular statement. For example:

while ( x <100 or y < 200 ) { x++; y++; }

2.2.4 Component Property Batch Setting

You can set multiple properties of a component in one statement, as below:

astruct.(prp1=1, prp2=2, prp3=3);

2.2.5 Comments

To put comments into Luban code is the same as C++. Anything from // to the end of line are comments. Anything between a /* to a */ are comments too. For example:

; // This line does nothing

; /* this line does nothing either */

2.3 Luban Built-In Data Types

Luban has a set of native data types to support most common operations. Below describes the native data types and their relations.

2.3.1 Double and Integer:

The key words for double (double precision floating point number) and integer types are double and int in Luban. Their relationship is similar to the double and int type in C. As below examples:

10/3; // result is integer 3

10.0/3.0; // result is double 3.333333…

10/3.0; // result is double 3.3333… mix double and int, result in double

0 == 0.0; // result is true, you can mix and compare double and int

2 > 0.1; // result is true, you can mix and compare double and int

x=1.23456789;

y=x.format(“.4”); // y= “1.2346” format to 4^th decimal place

z = int(x); // z is 1, cast double into integer

xstring = “1.23456E7”;

xdouble = double(xstring); // x = 1.23456x10^7 construct double from string

2.3.2 Boolean

The key word for Boolean type is bool. Boolean can only be true or false. And unlike C, There is no implicit casting to and from any other type. And the operators for Boolean type data are not, and, or. Applying arithmetic operation to Boolean type will result in error. The condition expression used in IF, WHILE and FOR statements must result in Boolean too, or the Luban code execution will stop.

One more thing to mention is that if you like, you can still use the C style &&, || and ! operators in place of and, or, not.

2.3.3 Character

Character type key word is char. It contains a single character. Below code example shows some operations on character type.

for( onechar =’a’; onechar <= ‘z’; ++onechar)

std::print(obj=onechar);

Above Luban program prints out “abcdefghijklmnopqrstuvwxyz” on screen..

NinetySeven = int (‘a’); // NinetySeven = 97, char to ASCII

Lettera = char(97); // Lettera=’a’, ASCII to char

2.3.4 String

String type key word is string. The below example Luban code shows operations for string type.

s = “Hello”; c = s[1]; // result is ‘e’

s = “Hello”; s[0] = ‘h’; // s becomes “hello”

s = “Hello”; foreach( c in s ) std::console().write( c ); // will print out Hello

s = “Hello”*2; // result is “HelloHello”

s = -“Hello”; // result is “olleH”

s = “Hello” + “World”; // result “HelloWorld”

// below are member function examples

// SPLIT !!

s = “Hello World”; words = s.split( ); // words is a vector [“Hello”, “World”]

// trim white space from front and back

s = “ Hello”;

s.trimfront(); // s becomes “Hello”

s = “Hello \n\t”;

s.trimback(); // s become “Hello”

//search

s=”Hello”;

index=s.find(“Hell”); // index = 0

index2=s.find(“Hell”,2); // index = null, faled to find Hell starting from index 2l

index3=s.find(“Heaven”); // index3 = null failed to find

truth=s.contains(“Hell”); // truth = true

sub=s.substr(2,3); // sub=”ll” from index 2 to index 3

// content manipulation

s=”Hello”;

s.replace(“Hell”, “Heaven”); // now s becomes “Heaveno”

s.lower(); // s becomes “heaveno”;

s.upper(); // s = “HEAVENO”

s.remove(6); // s=”HEAVEN”, with 0 at index 6 removed

s.remove(“A”); // s=”HEVEN”;

s.insert(1, “A”); // s=”HEAVEN”;

s.clear(); // s=””

2.3.5 Vector, Map and Set

Vector type key word is vector.

x = 1; y=2; z=3; v = [x,y,z,4]; // v = [ 1,2,3,4]

v[0] = 10; // v = [10,2,3,4]

total = 0; foreach( element in v ) total += element; // total =

v.sort(); // v become [2,3,4,10]

v2 = v+ [11,12,13]; // v2 become [2,3,4,10,11,12,13]

v2.remove(0); // v2 become [3,4,10,11,12,13]

v2.removeobj(10); // v2 become [3,4,11,12,13]

v3 = [ v2,v2]; // v3 is two dimensional vector

// push and pop

x=[1,2];

x.pushfirst(0); // x =[0,1,2]

x.pushlast(3); // x=[0,1,2,3]

y=x.popfirst(); // y=0, x=[1,2,3];

z=x.poplast(); // y=3, x=[1,2]

Map type key word is map. Keys are unique in a map. Code examples:

x=1; y=2; z = 3; word2number = { “one” : x, “two” : y, “three” : z };

word2number[“four”] = 4; // insert or replace

word2nubmer.remove(“one”); // remove key “one” and associated 1

word2number.insert(“two”,2.0); // no impact because key already exist

union = word2number + { “five”:5, “six”:6};

foreach(key,val in word2number ) std::console().writeline(key,’ ‘,val); // iterate

emptymap = {:};

Set key word is set. Set is simply a collection of unique values. See examples:

oneset = { 1, “hello”, 3.14 };

another = { 1,2};

union = oneset + another; // union = { 1,2,”hello”,3.14}

minus = oneset – another ; // minus = { “hello”, 3.14}, common values removed

joint = oneset – minus; // joint = {1}, the common values of oneset and another

foreach( val in oneset ) std::console().writeline(val); // iterate

emptyset = {};

Casting operations among vector, set and map:

oneset = { 1,2,3}; onevec = vector(oneset);

keyvec=[‘a’,’b’]; valvec = [‘A’,’B’]; onemap = map(keyvec, valvec);

2.4 Common Operations For All Luban Types

All the common operations like +-*/%, [] (indexing), member function call and type casting can be applied on all Luban data types, though foreach statement can be applied on all Luban container data types. Though the actual results of the operations may differ depending on the data objects in the operation. There could be the case the particular operator does not apply to the data object, and the operation will result in error value in that case. It is part of Luban’s polymorphism. The same operation results differently for different data types.

2.4.1 Arithmetic Alike Operation Examples

// A+B operations

numtwo = 1 + 1; // numtwo = 2

stroneone = “1” + “1”; // stroneone = “11”

vec = [1]+[1]; // vec = [ 1, 1]

s = { 1 } + { 1 }; // s = {1}, set merge

m = { 1:10 }+{ 2:20}; // m = { 1:10, 2:20 }, map merge

err = “1” + 1; // err = error, can not add string and integer

err2 = true + false; // err = error can not add logical values

//A – B operations

zero = 1 – 1; // zero = 0

varset = { 1,2,3} – { 3,4,5 }; //varset = {1,2} remove common elements of two sets

err = [1]-[1]; // err = error, can not subtract vectors

// -X negation

x = -100; // x = -100

vec = - [ 1,2,3]; // vec =[3,2,1] vector negation reverses element order

str = - “abc”; // str = “cba” string negation is order reversing

truth = -false; // truth = true logical value negation equals NOT operation

err = - ‘a’; // err = error, can not negate char type

// A*B operations

x = 2*2; // x = 4

str = “ab”*2; // str = “abab” string multiplied by integer

vec = [1,2]*2; // vec = [1,2,1,2] vector multiplied by integer

// A/B operations

x = 100/3; // x = 33 integer division

xfloat = 100.0/3.0; // xfloat = 33.3333…. float number

// A%B

xmod = 100 % 3 ; // xmod = 1

// ++ operation

x = 0;

y = ++x; // x=1, y=1, x goes up by one first, then assign to y

z = y++; // z =1; y =2; y assign to z first then goes up by one

x = ‘a’;

++x; // x= ‘b’ next of ‘a’ is ‘b’

-- operator is actually the reversion of ++ operator. Below are examples for operation-and-assign operators.

x += 10; // same as x = x+10;

x -= 10; // same as x = x-10;

x /= 10; // same as x = x/10;

x *= 10; // same as x = x*10;

x %= 10; // same as x = x%10;

It is straight forward that operation-and-assign operator is faster than the equivalent two-step operation then assign operation.

2.4.2 foreach Statement And Container Type

foreach statement only applies to container types including vector, map, set and string types. Applying it on non-container type will result in the termination of evaluation of the current Luban component.

X = 3.1415;

foreach( y in x ) // ERROR! Terminate here

std::println(obj=y);

Below is an example of correct use of foreach statement.

foreach( x in [10,20,30] ) std::println(obj=x);

Above code will print 10,20,30, one number per line.

2.4.3 Common Member Function Calls

All Luban data object has member functions. The member functions for each data object are defined by the type of that data object. Though there are a collection of common member functions available for all Luban data types, including both built-in and user defined types. The below is the list of those common member functions and their descriptions.

H= x.hash(); // hash key for the object, default value 0

S= x.size(); // apply to container and string type, default value error

Y = x.contains(y); // apply to string and container type

x.clear(); // apply to string and container

guts = x.serialize(); // returns the binary guts stream in a string

h = x.invoke(“hash”); // call member function hash in a reflective way

funcmap = x.allfuncs(); // map containing all member functions and descriptions

obj = std::des(stream=guts).obj; // restore object from guts string

The above last line is not a member function call. It just shows how to use the string generated from serialize() function call. All member function calls will return error value when error happens in the call.

2.5 Null and Error

Null is a special value in Luban, its key word is null. Null can be used in expressions just like a normal constant. The null value is special in the following two ways. First the null value is the default setting for all variables and properties. Second the null value is of no type.

Error type is a special type in Luban, with key word error. Value of error type can be generated by Luban internal operations, or explicitly constructed by Luban code. All member function calls return error value if things go wrong inside the function. For example:

errval = 1/0;

if ( not errval isa error )

std::console().writeline(“Your Luban engine is on fire”);

expliciterr = error(“Hey, I am a born error”);

2.6 Parallel Computing: Threading and Synchronization

Multi-threading is a built-in part of Luban, as the below sample code shows.

myjobs::dothis(arg=100) & // dispatch the expression in a different thread

myjobs::dothat(arg =200) & // dispatch thread

{ c.doOne(); d.doTwo(); } & // dispatch a block

wait; // wait for all three threads to finish

You can selectively wait for certain thread to finish, as far as the thread can be associated with a variable. For example:

x = y.domagic() & a = b.dogood() & // dispatch two threads

waitfor x; // wait for first thread to finish

{ x = y.domagic(); a = b.dogood(); } &

{ c = d.doevil(); f = g.wastetime(); } &

waitfor f; // wait for the second thread to finish, the thread is identifiable by the last assignment variable

Luban engine guarantees the object integrity in the multi-threading environment. If more than one threads operating on the same object, the order of operation is not deterministic, though the integrity of each individual operation is guaranteed. Luban maintains object integrity by allowing only one thread to operate on a variable at any time of execution. Though if user builds a data type internally doing pointer reference counting, then user is responsible to make this data type thread safe. Because Luban can only see different variables and there is no way for Luban to be aware of the internal data sharing of the user data type.

One question is what will happen to those dispatched threads when the main execution thread comes to an end. By default Luban will wait for every dispatched thread to finish before doing the cleaning up. So it is like there is a hidden wait statement at the end of all scripts and structures. You can change this behavior by explicitly using cancel statement.

2.7 Tangled Threads Surprise

Take a look at this example:

for(i=0;i<10000:++i) { x += i; } &

The above seemingly innocent code will almost certain fail because it will try to start 10000 threads that will likely be more than any machine can handle. Above code will start one thread per loop. The right way to put a loop into a thread is to whole loop into a pair of brace brackets like below:

{ for(i=0;i<10000:++i) { x += i; } } &

Thread can be tricky sometimes. For example, the below simple code can produce output that surprise you.

for( i=1; i<=10; ++i )

{

std::println(obj=i) &

}

At the first look, you would think the program print out number 1 to 10 in that order. But the truth is that the program will print out 10 numbers, but which 10 numbers and the order of the numbers are undetermined. The print could vary at each run of the program.

The root of chaos is the thread dispatching part of this code. Every print operation will be dispatched to a different thread. So ten threads will be dispatched. Together the main thread, there will be eleven threads running in this code. The execution order of these eleven threads is not deterministic.

You can use the wait and waitfor statements to control the thread execution. Plus there is another tool can be used to tame the chaos. The tool is called Asynchronous Structure. And it will be described later in chapter seven.

2.8 Everything Is Object

On handling variables and objects, the major difference between Luban and other major languages like C++ and Java, is that all objects and variables in Luban are passing by value. There is no reference or pointer type in Luban. When assigning one variable to another in Luban, the object is semantically copied instead of passing a reference like in Java. Actually Luban’s value semantics is very similar to C, except Luban does not have the pointer type. For example:

a = [1,2,3];

b = a;

b = b + [4,5]; // b = [1,2,3,4,5] and a = [1,2,3]

The reason behind the design of Luban’s value semantics is simplicity. Value semantics is straightforward and intuitive, requiring almost zero learning to understand.

3 Introduction to Luban Component Structure

The least flexible component of any system is the user.

- Lowell Jay Arthur

This chapter introduces the basics of Luban component model. The foundation stone of Luban component model is the Luban structure, which is indicated by a key word struct in Luban. The below will show how to define and use a Luban structure.

3.1 Define a Luban Structure

Suppose our hello-world program becomes so popular that we want to reuse it as a component. The way to do it is to define the hello-world code as a Luban structure, as below example shows:

namespace demo;

struct helloworld( )

as process

{

std::console().writeline(“Hello, world!”);

}

Now let’s save the above code as file “hello.lbn”. Then we can write the below script to call this component.

demo::helloworld(=);

We then save this script as “callhello.lbn”. And finally we can run this new componentized hello-world program by typing the following command in the command window of your operating system:

Mycomputer > luban hello.lbn callhello.lbn

Hello, world!

Let’s take a look at this new hello-world program in details. In “hello.lbn” file, we define a Luban structure named “helloworld”. Luban requires every component must reside in a name space. That is why we see the line “namespace demo;” at the top. So the full name of this component is “demo::helloworld”. The key words “as process” indicate the body of the Luban structure is a process. The difference between a process structure and other structure will be discussed in later chapters. All we need to know now is that we can use all the constructions used for scripting in the body of a process structure.

In the next script “callhello.lbn”. We make a call to the component by using its full name. The syntax “(=)” indicates a dynamic call to a Luban structure. The details of the dynamic call to a Luban structure can be found in later chapter and here we only need to know the call will invoke the execution of the code in the Luban structure “demo::helloworld”.

One thing need to be mentioned here is that in the above example Luban source code, there is no referring to the actual source code file name, component is only referred by its full name. This is no incidental, it is purposely designed so to avoid hard coding source code media. Yet we don’t care about this for now.

3.2 Hello World Structure with Input and Output

The previous example defines a Luban component structure without any input and output properties. In practice, a component as a processing unit may need input and output properties to pass information, as the below example shows:

namespace demo;

struct greeting

(

input:

saywhat, towhom;

output:

greetingmsg;

)

as process

{

msg = input.saywhat + “, “ + input.towhom + ‘!’;

std::console().writeline(msg);

output.greetingmsg = msg;

}

The above Luban code defines a Luban structure with two input properties, saywhat and towhom. The internal process of the structure combine the two input property values into a single greeting message and set it to output property named greetingmsg. The side effect of this structure is to print out the greeting message to the console. As we can see, inside the structure, “input.property” is the convention used to refer input and output properties.

As you can see from the example, a Luban structure definition has two parts. First is the structure interface definition, which is defined by listing the input and output properties in between the parenthesis after the structure name. The second part is the internal implementation of the structure. In this the internal of the structure is defined as a process with code between the brackets.

Now let’s save this Luban component code as “helloio.lbn”. Then we need to write a script to use this new component, as below:

demo::greeting(saywhat=”Hello”, towhom=”world”);

Save the above one line in a file named “callhelloio.lbn”, then run the following command on your operating system like below:

Mycomputer > luban helloio.lbn callhelloio.lbn

Hello, world!

From the example, arguments are passed into structure to be called by using propertyname=value pairs in parenthesis. The syntax difference between a Luban structure call and regular function call in C is that instead of lining up the arguments by order, arguments are explicitly paired with input properties by name.

3.3 More On Use of Luban Structure

So far the previous examples show the use of Luban structure in a way just like calling functions. But there is a key difference between C function and Luban structure. Luban structure is object with state indicated by its property values, while C function is mostly stateless. Luban structure is more flexible and can be used in many different ways, as below example shows:

greeter = demo::greeting(saywhat=””, towhom=””); // construct

greeter.saywhat = “Hello”; // print Hello,!

greeter.towhom = “world”; // print Hello, world!

greeter(saywhat=”Nice to meet you”); // print Nice to meet you, world!

greeter(=); // print Hello, world!

greeter.(saywhat = “How are you”, towhom=”Luban”); // print greeter(=); // print How are you,Luban!

std::console().writeline(greeter.greetingmsg);// print How are you, Luban!

Save this file as “callhelloalt.lbn”, then run the script as below:

Mycomputer > luban helloio.lbn callhelloalt.lbn

Hello,!

Hello,world!

Nice to meet you,world!

Hello,world!

How are you, Luban!

Here we explain some details of the script line by line. The first line construct a Luban structure, assign to a variable, while triggering the execution of the structure once in the process. That is why we see a “,!” output line above. The next line set the value “Hello” into the input property named “saywhat” of the structure in variable “greeter” and the setting of input trigger the execution of the structure. That’s why we see the second output line “Hello,!”. Similar process happens for the third line of code, which sets the property “towhom” to value “world”. The fourth line is a structure call to a structure variable instead of to a structure type name as we previously know. And this call overrides the property “saywhat” with value “Nice to meet you”, making the fourth output line “Nice to meet you, world!”.

It is important to know that the call to a structure does NOT change the state of the structure at all. Even with the explicit overriding of property values in the call, the property values of the original structure remain the same during the call. Actually the structure call generates a temporarily new structure that can be assigned to a different variable. And that explains the fifth line of code calling the original structure without arguments ends up printing the same greeting as before.

As mentioned above, property setting will trigger the execution of the structure. Sometimes we may want to set multiple properties while we only want one code execution after we finish. That is why we have the sixth line, which setting multiple properties of the structure to different values, and executing the code once the settings of properties finish. And that is why we see the sixth output line “How are you, Luban!”. Just to show the setting of multiple properties does change the state of the structure, the seventh line call the structure again without arguments and see the “How are you, Luban” output once again. The last line of the script read the output property “greetingmsg” of the structure and print it to console. That’s the third “How are you, Luban!” from.

3.4 Pass By Value Semantics

It is necessary to once again iterate the pass by value semantics of Luban. The passing of arguments in Luban structure call is by value, plus it is a one way street. What happens inside the structure has no possible impact on the caller. For example, the following code will result in syntax error:

namespace demo;

struct swapobj

(

input:

x,y;

)

as process

{

temp = input.x;

input.x = input.y; // this line has syntax error

input.y = temp;

}

The reason for the syntax error is that Luban does NOT allow assignment toward input property inside a structure. Input properties are read only inside the structure.

3.5 Name Space Basics And Program Entry Point

There are two kinds of Luban source code unit ( i.e file ) that Luban interpreter accepts. One is Luban script, which is a sequence of Luban process statements and expressions. Luban interpreter will simply execute the script upon request.

Another kind of source code unit is component structure definition module that defines Luban structures. The first line of a component definition module must be the definition of name space in which the following structures will reside. Luban name space uses a syntax similar to C++ name space. When Luban interpreter encounter a component definition module, it will scan the module and register the structure components into a global name space data structure so it can be used in later call. Luban script can be used as the entry point of an application or system, while Luban component definition module can be used like library.

4 Structure Interface and Inheritance

Different towns, different people,

Somehow they're all the same.

- Long Time Gone, Bob Dylan

Component oriented programming can be compared with LEGO toys in many aspects. First you need to put the parts into the right boxes so you can easily find them out when you need them. That is what Luban name space does: put components into organized places. Second you need to recognize the parts by their shape and the way they make connections so you can use them in the right place. And that is what Luban structure interface and inheritance do: formalizing the contract with external world by defining interface, and reflecting the similarity among different components by inheritance. The goal of Luban name space, structure interface and inheritance is the same: to keep our component workshop nice and neat.

4.1 Structure Interface: Property Flow Type and External Access Permission

As shown by the examples in previous chapter, the interface of Luban structure consists of a list of different kind of properties. The previous examples have shown the basic usage of input and output properties. And below we describe all different kind of properties and their behavior in a little formal way.

There are four different kinds of flow types for properties in Luban: input, output, store and static. And the properties can have different read/write permission for external depending on their flow type. There are three kind of permissions: readwrite, readonly and hidden, and using these key words can only change the access permission for external caller. From within the structure the access permission for properties are predefined and unchangeable. The purpose of this design is to define a general framework of information flow for Luban components. The below table shows the details:

input: For external, readwrite. For internal, read only. Both unchangeable.

Change of value triggers evaluation of the structure.

output: For external, readonly. For internal, write only. Both unchangeable.

Value can only be changed from internal by evaluation.

store: For external, hidden by default, changeable to readwrite or readonly

For internal, both read and write.

Change of value does not trigger evaluation.

static: For external, readwrite by default, changeable to hidden or readonly

For internal, both read and write.

Change of value does not trigger evaluation.

Accessible only through structure type name.

From the above table and previous examples, we can see that for a Luban structure data flows from input to output. And the flow can no way be reversed. This flow model is as fundamental for Luban as function semantics for C.

We have known input and output properties from examples. The below example shows the use of store and static properties:

namespace demo;

struct aggregator

(

input incoming;

output total;

store sofar=0;

)

as process

{

store.sofar = store.sofar + input.incoming;

output.total = store.sofar;

}

struct useagecounter

(

static readonly totaleval=0;

)

as process

{

static.totaleval++;

}

Let’s save above code into a file “storenstatic.lbn”. Then we can test our new structures with the below script:

// testing aggregator

agg = demo::aggregator();

for(i=1; i<=100;i++) agg.incoming = i;

std::console().writeline(agg.total);

// test static

counter1 = demo::useagecounter();

counter2 = demo::useagecounter();

counter1(=); counter2(=);

std::console().writeline(demo::useagecounter.totaleval);

Then we save the above script as file “testsns.lbn”. Afterwards we run the test by typing the following in the command window:

Mycomputer > luban storenstatic.lbn testsns.lbn

5050

The structure “aggregator” adds all numbers setting to its “incoming” property together and output the total to “total” property. The store property “sofar” in aggregator is used for internal state keeping. Because Luban prohibits reading value back from output properties, store property is used when some state data need to be remembered among different evaluations. Store property is similar to private data of C++ class.

The structure “useagecounter” maintains a global static property “totaleval” to count the number of evaluations of the structure. The static property is actually associated with structure type instead of structure instance. From outside the structure, the static property can only be referred by <structure type name> .<static property name>. From within the structure, it can be accessed using internal property naming convention that is <flow type>.<property name>.

4.2 Structure Interface Inheritance

The interface is the service contract of structure with the external world. We have seen the property flow types and external access permission key words, which are basic terms that can be used to draft such a contract.

The structure interface declares what each individual structure type does. Though when the number of different structure type grows, a large collection of unrelated structure interfaces are still difficult to remember and manage. And that is where the Luban structure interface inheritance come to help.

Similar to the class inheritance concept in Object-Oriented programming, we want to take the advantage of commonalities among different components by categorizing them using inheritance. The idea is that it is easier to understand ten animal kind than to understand one hundred individual strange animals. The below example shows how to code inheritance in Luban:

namespace demo;

struct DualOp

(

input:

op1,op2;

output:

result;

)

as process

{

output.result = error(“No actually operation for general dual operand”);

}

struct Adder implements ::DualOp

(

)

as process

{

output.result = input.op1 + input.op2;

}

The above Luban code defines two structures. Structure “DualOp” is a general interface for two operand operations. Structure “Adder” inherits its interface from “DualOp” and implements the internal to add two inputs together and set to output.

You may notice the syntax “::DualOp”, which is a short name for “demo::DualOp”. All Luban’s structure type names are full name. The short name can be used if the name space of the referred Luban type is the same as the name space of the caller component.

From this example, we can see that “Adder” inherits all input and output properties from “DualOp” and does not define any new property of its own. And we need to mention here that ONLY input and output property can be inherited. The store and static property can NOT be inherited. The reason for this design is to make interface inheritance simple.

4.3 Luban Inheritance Is Only For Interface

Unlike the general Object-Oriented languages like Java and C++, Luban’s inheritance mechanism is only for the interface. When one Luban structure inherits from another, only the interface of the parent structure is taken, and that excludes store and static properties. If you wish to use the internal implementation of the parent structure, you need to express you intention explicitly, as below example shows:

// this example uses the Adder structure defined in previous example

namespace demo;

struct AdderWithUseCount implements ::Adder

(

static usecount=0;

)

as process

{

static.usecount++;

output = ::Adder(input=::Adder.input).output;

}

The above example defines a structure “AdderWithUseCount” that works the same as the structure “Adder”, only with one more static usage counter property to count the number of evaluations.

The magic in this example is the last line of the structure process code. This line simply says: take the portion of input inherited from “Adder” structure, feed to a call to structure “Adder”, then take the output of this call and set to the output of structure “AdderWithUseCount”.

What actually happens in this example is very much like AdderWithUseCount inherits both interface and implementation from Adder. Though the taking of “Adder” structure’s implementation is explicitly stated in one line of code. The purpose of this design is to avoid the tricky ambiguity issue for inheritance by allowing programmer to state intention explicitly. Luban allows multiple inheritance of structure interfaces.

4.4 Abstract Structure

Similar to interface concept of Java and pure virtual class of C++, Luban has abstract structure to represent pure structure interface. Abstract structure can be used for inheritance and type checking. We can redo our “Adder” example using abstract structure like below.

namespace demo;

abstract struct DualOp

(

input:

op1,op2;

output:

result;

);

struct Adder implements ::DualOp

(

)

as process

{

output.result = input.op1 + input.op2;

}

In above sample code, we define an abstract structure named demo::DualOp and demo::Adder inherits its interface and implements actual operation. The inheritance relationship can be checked with type checking expression, like below Luban code shows:

adder = demo::Adder();

test = adder isa demo::DualOp;

The value of the variable test in above example is Boolean value true. The details of type checking expression can be found in next chapter.

Abstract structure can only be used for inheritance and type checking. Construction of abstract structure not allowed. For example, below Luban expression results in error value.

errvalue = demo::DualOp();

4.5 No Cyclic Inheritance

Cyclic inheritance is not allowed in Luban. Not only direct cyclic inheritance among structure interfaces is not allowed. All structure type names used inside structure interface definition including names used in property data type definition and property value initialization expression are considered part of its dependent symbols. Luban requires all its dependent symbols (interface ) resolved before the structure interface can be resolved itself. Thus cyclic reference among dependent symbols will also result in error. For example, below Luban structure definitions are invalid because of cyclic reference of structure type symbols.

namespace demo;

struct TweedleDee

(

input:

x = ::TweedleDum();

output:

)

as process

{

output.y = 1.0;

}

struct TweedleDum

(

input:

a = ::TweedleDee();

output:

)

as process

{

output.b = 1.0;

}

The above two structure “TweedleDee” and “TweedleDum”, though they don’t directly inherit from each other, they refer to each other in their property initialization expressions. And that is also cyclic interface definition and it is invalid.

4.6 Stationary Structure

Stationary Structure in Luban is used purely to hold data. It can be used almost the same way as a generic map data object. But the difference with Stationary Structure and map is that Stationary Structure is typed (so type checking works for it) and it is much faster than generic map data type. It is a good practice to define always bundled data stream as Stationary Structure to make it standard, similar to data bus concept in hardware design. Below sample code shows how to use Stationary Structure.

namespace demo;

stationary struct GlobalStation

(

static hostname, cpuinfo, osinfo;

);

stationary struct DataPair

(

name, value;

);

Above code defines two Stationary Structure, one to hold static global information, another one to collect two always paired data item into a single structure.

By default the properties defined in Stationary Structure always have their access permission as readwrite unless otherwise explicitly specified.

Stationary Structure can be inherited, serialized and de-serialized the same way as regular structures.

5 The Dynamic Type System of Luban

Life is a box of chocolate,

You never know what you’re gonna get

- Forest Gump

As we all agree with Forest Gump about life’s unpredictability, we may cope with it by checking each chocolate before we take a bite. And that is exactly what Luban’s dynamic type system does. An object can be anything, yet you can check what it is if you want to.

All the objects inside Luban can be separated into two categories: data types and structure types. Structure types are the components coded in Luban. You can create as many kinds of Luban structure types as you wish just like you can code as many functions as you wish in C.

Data types are those built-in and/or imported data types with predefined set of member functions and applicable operators. Map, set, vector, double, int, bool, char and string are the examples of built-in data types, while std::file, net::socket are examples of imported C++ data types. The fundamental difference between structure type and data type is that you can not code a new data type with Luban. One design philosophy of Luban is that the basic data types are low level software elements that should be accessible from Luban, but Luban should not be used to make low level elements. Just like bricks are used to build house, instead of bigger bricks. Luban’s belief is there should be a fundamental programming paradigm shift from software element level to component level.

Though structure types and data types are naturally different. They still obey a more or less common set of rules. These rules are to be discussed in this chapter.

5.1 Name of Data Type and Structure Type

The naming rule of Luban data type and structure type are the same. They all must be a full name with name space. But Luban built-in types are keywords and do not need name space prefix. For example:

string; // built-in type

std::file; // standard imported data type

net::listener; // imported data type

net::socket; // imported data type

std::print; // a structure type

The name space part in a type name can be omitted only when referred in a component that resides in the same name space, as below code shows:

namespace std;

struct print

(

input obj;

)

as process

{

::console().write(input.obj);

}

The above code simply wraps the calling of member function of std::console type in a structure named std::print. We can see the referring to std::console in std::print structure does not use the name space part.

5.2 Construction of Data Type and Structure Type

The syntax of object construction in Luban is type name followed by parenthesis, as shown in the below example:

// print hello world

console = std::console();

console.writeline(“Hello, world!”);

// print hello world again

printer = std::print();

printer.obj = “Hello, world!\n”;

Parameters can be used in the object construction. But the syntax of argument passing is different for data type and structure type. Look at the below example:

// create file as data type

file = std::file(“helloworld”, ‘w’);

file.writeline(“Hello, world!”);

// print hello world to console

printer = std::print(obj = “Hello, world!\n”);

To pass parameters into data type construction, you just need to line the parameter values up. To construct a structure with specified property values, you need to pair up the property names with its values. Another big difference for structure construction with specified property values is that if the specified properties include input property, the structure will be evaluated once after construction. Yet the structure construction without any specified property values does not trigger any evaluation.

5.3 Explicit Type Casting

Type casting in Luban is just an object construction with a single argument. Like below:

int(“100”); // result 100

string(100); // result “100”

double(100); // result 100.0

int(3.1415); // result 3

Casting to string type is special because it is universally enabled for all Luban built-in and imported data types. Actually when std::console type is asked to print an object, it will cast the object to a string then print the string.

Type casting does not apply to structure type.

5.4 Dynamic Type Checking: isa and typeof Operators

You can dynamically check the type of any object in Luban with isa and typeof operators, as shown in below Luban script:

d = 3.1416;

s = std::file(“pie”, ‘w’);

// check type and decide what to do

if ( d isa double and s isa std::file )

s.writeline(d);

else

s.writeline(“The end of world is coming”);

// do the same thing in a different way

if ( typeof(d) == double and typeof(s) == std::file )

s.writeline(d);

else

s.writeline(“The end of world is coming”);

The structure interface inheritance is taken into account for type checking. For example the variable “test” in below Luban code has value true because structure demo::Adder inherits interface from demo::DualOp ( Previous example):

adder = demo::Adder();

test = adder isa demo::DualOp; // test = true

5.5 Type Expression

One unique feature of Luban is type expression, which gives more power and flexibility to Luban’s type system. Type expression is part of the Luban expression. The result of type expression can be assigned to variable and passed around just like normal object. The details of type expression can be shown in the following Luban code examples:

Type name alone can be a valid type expression.

stype = string; // type name is type expression

if ( not “Hello” isa dtype ) std::console().writeline(“Insane”);

Multi-type type expression is a collection of different type expressions. The isa operator returns true if the type of object matches one of the types in a multi-type.

streamtype = < std::file, net::socket >

s1 = std::file(“destfile”,’w’);

s2 = net::socket(“destserver”, 6500);

if ( s1 isa streamtype and s2 isa streamtype ) std::println(obj = “Good”);

Enumerated type is a collection of constants. The isa operator returns true if the value of object equals any one of the collection. Look at the below example:

littleprime = < 1,2,3,5,7 >;

test = 5 isa littleprime; // test = true

The type name for type object is typeinfo. Below shows how it works:

stringtype = string;

test = stringtype isa typeinfo; // test = true

5.6 Gate Keeper: Typed Property and Variable

There are tools in Luban to enforce strong typing policy if necessary. The type of property can be explicitly declared with type expression in structure interface definition to make sure only the objects of specified type can pass through. And also local variable can be bound to certain type using type expression. The setting of object of wrong type to a typed variable will result in the abortion of structure evaluation or script execution.

We can see the use of typed variable from below Luban code:

double dx;

dx = “Hello”; // evaluation aborts here

std::console().writeline(“You should never see me, the world should have ended before me”);

The above example assigns a string value to a variable “dx” that is a variable with predefined type double. The assignment results in the aborting of the evaluation of the script. So the last line of code should never be executed and the print should never appear on screen.

The below example shows the use of typed property in structure interface:

namespace demo;

struct powertoint

(

input:

< double, int > base;

int power;

output:

result;

)

as process

{

result = 1;

for( I = 0; I < input.power; I++)

result *= base;

output.result = result;

}

We can save the above code to a file “powerint.lbn”. Then we can write another script to call this new component, as below

sqr = demo::powertoint( base = 4, power = 2 ).result; // pwr = 4*4

std::println(obj=sqr);

sqrt = demo::powertoint( base = 4, power = 0.5 ).result;

if ( sqrt isa error ) std::println(obj=”Good, to get square root failed as expected”);

pwrcalc = demo::powertoint();

pwrcalc.power = 0.5; // evaluation stops here

std::println(obj=”Surprise! World should have ended before me”);

We then save the above script as “testpowerint.lbn” and start the test like the following:

Mycomputer > luban powerint.lbn testpowerint.lbn

Good, to get square root failed as expected

We can see the first evaluation of demo::powertoint succeeded and gave result 16 to print. The second structure call having property “power” set to 0.5 that violates the int type of the property results in an error value, causing the printing of “square root failed” message. The line before the last line sets a double value into the property “power” of int type, causing the evaluation to stop. That’s why we don’t see the last line get executed.

5.7 Typedef to Save You Some Typing

You can use typedef to define a symbol that is just the alias of another symbol. You can save some typing or create a layer of encapsulation in this way, as below code shows.

namespace demo;

typedef std::print print;

Now you have a symbol “demo::print” that is an alias to the real type “std::print”.

You can also use typedef to define commonly used type expression as a symbol, like below:

typedef <std::file, net::socket, std::console> media;

Now you have a type symbol “media” that represents one of the three types.

6 Component Composition

Poor boy, pickin' up sticks

Build ya a house out of mortar and bricks

- Po’ Boy, Bob Dylan

So far we have learnt how to script Luban and write Luban structure component as process code. As we mentioned before, there is another way to define a Luban structure, which is composition. The way composition defines a Luban structure is to lay out the sub-components and specify the dependencies among them. The evaluation mechanism of composition cell is to propagate the data change The Luban structure defined as composition has the same interface as process structure and can be used the same way. In this chapter, we go through the component composition with a series of examples.

6.1 Composition 101

The basic format of Luban composition is simply a list of sub-components. The sub-component in a composition is called component cell. Each component cell has a name and a definition. The definition of the component cell specifies the content of the component cell and its dependencies with other component cells outside. The below example illustrate a basic composition:

namespace demo;

struct simplecomp ( output oneoone; )

as composition

{

A1: 100;

A2: A1+1;

output.oneoone: A2;

}

The above composition computes 100+1 in a way like spreadsheet. There are two component cells in above structure, A1 and A2. A1 contains a number 100. A2’s content is an expression “A1+1”. The last line links the result of A2 cell with output property “oneoone”. Then we use the below Luban script to call this structure:

result = demo::simplecomp(=).oneoone; // result = 101

To show the difference between composition and process, we change the structure “simplecomp” to below:

namespace demo;

struct simplecomp ( output oneoone; )

as composition

{

output.oneoone: A2;

A2: A1+1;

A1: 100;

}

The structure “simplecomp” behaves exactly the same as before. In Luban composition, the order of listing of the component cells does not matter, as far as all the cells and their dependencies remain the same, unlike the process structure for which the order of operations is essential.

6.2 Different Component Cells

There are two kinds of component cells that can be used in a Luban composition, ad hoc cell and typed cell. We discuss these two kinds of cells in this section.

6.2.1 Ad hoc Cell

Ad hoc cell contents are very much like what people normally put into spreadsheet cells. They can be constants, expressions. In Luban composition, the ad hoc cell can also contain Luban script code to do more sophisticated processing, as shown below:

namespace demo;

struct compadhoc

(

input in1, in2;

output out;

)

as composition

{

A1: input.in1 + 1;

A2: input.in2 + 1;

A3: { std::console().writeline(“run A3”); A3=A1*A2; }

A4: { std::console().writeline(“run A4”); A4=A2*A2; }

output.out: A3+A4;

}

The above composition uses only ad hoc cells. The cell A1 and A2 are simple expressions. Just as expected, Luban will evaluate the expression and save the result in the cell.

Cell A3 and A4 contains Luban script code. When data updating event reaches these two cells, Luban execute the script code just like normal script. The difference between an expression cell and a script cell is that script cell does not get any value returned from script by default unless the script in the cell explicitly set the value of the cell itself, like cell A3 and A4 do in above code.

The last line in above example links the result of expression “A3+A4” to the output property “out”. The output property linking cell is one special kind of cell that can only contains single expression, no script code is allowed.

6.2.2 Typed Cell

Typed cell is to enable user to directly put pre-defined structure component into composition, like below example:

namespace demo;

struct TwoAdder

(

input op1, op2;

output result;

)

as process

{

output.result = input.op1+input.op2;

}

struct FourAdder

(

input op1,op2,op3,op4;

output result;

)

as composition

{
adder1: demo::TwoAdder(op1=input.op1, op2=input.op2);

adder2: demo::TwoAdder(op1=input.op3, op2=input.op4);

adder3: demo::TwoAdder(op1=adder1.result, op2=adder2.result);

output.result: adder3.result;

}

The above example illustrates how to compose a adder that can add up four numbers from adder that adds two. You can see inside the “FourAdder” composition there are three “TwoAdder” components inside, two of them linking to input properties while the third one adds the result of first two and send result to output property.

Actually the typed cell is not much different from expression cell, except that the typed cell can not be used by its own value like ad hoc cell. The other cells of the same composition can only use the output property of the typed cell.

6.3 Dependency Specification Rules

Luban composition gives user the freedom to put expression, script and typed structure into component cells, while Luban figures out the dependencies automatically for you. Yet there are still rules for dependency specification to make sure the composition is semantically clear and unambiguous.

6.3.1 Change Nothing But Yourself, But Do Read Your Surrounding

In a composition, a cell can not modify other cells, it can only read from them. These operations are considered as modification: directly assignment, member function call and property setting. So if there is such operation upon other cells in a cell definition, Luban will report error. This is to make sure the data flows one way as commonly expected.

A cell can change itself by directly assign value to itself. Though a cell can not read the previous value of itself back. So calling member function on itself is also prohibited because it involve reading previous value back. The purpose of this rule is simply to avoid confusion.

6.3.2 No Tail Chasing

Luban does not allow cyclic dependencies among synchronized cells in a composition. The details of synchronized cells can be found in later chapter. So far in all the sample compositions we have seen, all the cells are synchronized.

6.4 Inheritance and Composition

Composition can be used together with Luban interface inheritance to create new structure based on old one, like below example shows:

namespace demo;

struct AdderPlus implements TwoAdder

(

output diff;

)

as composition

{

output: demo::TwoAdder(input=demo::TwoAdder.input);

output.diff: input.op2 – input.op1;

}

In above example, the structure “AdderPlus” takes the interface of “TwoAdder”. It also put one instance of “TwoAdder” into its composition and wired the right portion of its input to the “TwoAdder” and wired the output of “TwoAdder” directly to its output. So “AdderPlus” basically get interface and implementation of “TwoAdder” in this way. It also adds one extra output of its own, which is the difference of its two input properties. The first line in the composition can only be used with structure inheriting interface from other structure. Luban will do type checking to make sure the wiring work with inheritance. There could be ambiguity in the case of multiple inheritance. The basic rule to handle ambiguity is to explicitly specify the wiring of the output property on which the ambiguity occurs.

6.5 Composition VS Process

Composition does its best when there are significant number of components in a system and the dataflow among the components are complicated. It is difficult to code multi-directional dataflow with process that is basically one dimensional.

Process should be used when the system have sophisticated control logic, meaning if conditions and loops. In practice, it could be the case that small components more likely use process while large components could be more compositional.

7 Asynchronous Structure: Implicit Thread and Queue

Anyone who attempts to generate random numbers by deterministic means is,

of course, living in a state of sin.
- John von Neumann

So far the Luban structure we have seen can be called synchronous structure, meaning the property values of a structure reflect the state of the structure and you can expect the change of input properties will trigger the internal evaluation and likely the change of output properties to put the structure into a new consistent state. And the internal evaluation happens in an atomic fashion, meaning Luban guarantees that the structure is always in a consistent state in every access.

In this chapter we introduce a new kind of structure: asynchronous structure. Asynchronous structure can have all the input, output, store and static properties just like the synchronous structure. The major difference is the semantics of the set and get of input and output properties. For an asynchronous structure, the semantic of input and output is more loose. What is defined is that data flows into input and flows out of output, and there is no consistency pairing between input and output. And the data flows in and out a asynchronous structure will be queued, unlike the synchronous structure, the new property value will overwrite the previous one. The store and static property value of asynchronous structure will not be queued, queuing is only for input and output. Asynchronous structure behaves like pipe, while the synchronous structure is more functional.

The purpose of asynchronous structure is to model asynchronous components in system. The examples of asynchronous components include network messaging and interactive user interface. And usually large system more likely behaves in an asynchronous way.

In actual implementation of Luban, the asynchronous structure is implemented as an object with a live thread inside. The inside thread takes data from input queue and put the output to the output queue, while outside user can put new data onto input queue and take output from output queue. So asynchronous structure does multi-threading, only implicitly.

7.1 The First Asynchronous Structure Example

namespace demo;

asynch struct multiplexer

(

input flow1, flow2;

output merged;

)

as process

{

output.merged = input.flow1 &

output.merged = input.flow2 &

}

The above example defines a asynchronous structure named “multiplexer”. The structure merges data input from “flow1” and “flow2” into the output “merged” just like a telecom multiplexer. Inside the “multiplexer”, there are two user coded threads, one to read “flow1”, another to read “flow2”, both put the reading into “merged”. There are several details of asynchronous structure that make this structure work.

First, every read of input property removes one value from the input queue. When the queue is empty, the read operations will block and wait until new value comes in. Input property can not be read from outside. Remember in the case of synchronous structure, repeated reading of the same input property read back the same value.

Second, every assignment to the output property adds one value to the output queue. For the same operation, synchronous structure will only take the last assignment to the output property. Reading of output property from outside will block if the output property queue is empty.

Understanding the above details, we can code a simple case of producer and consumer in which producer and consumer share the same queue for communication.

namespace demo;

struct simplequeue

(

input inq;

output outq;

)

as process

{

outq = inq;

}

The above component behaves like a simple multi-thread queue with one end to put in and another end to read out data, plus it will make reader wait when the queue is empty. Then we can use the above component in a sample script like below:

// ……

q = demo::simplequeue();

{ obj= myns::producer().produce(); q.inq=obj; } & // producer thread

{ myns::consume( obj = q.outq ); } & // consumer thread

// ……

Above script starts two threads, one to produce object and put it into the “inq” of the queue named “q”, while another to read object out of “outq” of “q” and feed to “myns::consume” structure. When the “outq” is empty, the consumer thread will sleep.

Above examples show how to code asynchronous structure and use it in Luban script. The next section will show how to code and use asynchronous in composition.

7.2 Use Asynchronous Structure to Synchronize

In chapter one, when we introduce thread dispatching, we give out an example that shows some nasty surprise when threads are dispatched uncontrolled in a loop. Actually that kind of loop is commonly used in certain situation. For example you could write a server that watch a network port and dispatch thread to handle received message. And this server is simple a forever loop with thread dispatching inside, like below code shows.

while ( true )

{

x = port.getmsg();

::msghandler(msg=x) &

}

Above code is problematic, because the execution order of the main thread and dispatched thread is not deterministic. It could happen that the main loop goes faster that the message “x” is changed before the “::msghandler” thread read it. The way to fix this is to use Asynchronous Structure, as below code.

handler = ::asynchmsghandler();

while ( true )

{

x = port.getmsg();

handler.msg=x;

}

And you define your “asynchmsghandler” like below:

asynch struct asynchmsghandler(input msg;)

as process

{

// handle msg;

}

Since the object “handler” contains an Asynchronous Structure that has a thread inside. And all inputs set to “handler” will be correctly queued to be processed in order.

7.3 What Happens to Asynchronous Structure When Script Ends

In chapter one, we described that at the end of script execution, the main thread will wait for all explicitly dispatched threads to finish. Luban engine does not give any special treatment to the live thread inside Asynchronous Structure, meaning it just destroy it like anything else. And the destruction of Asynchronous Structure will stop the internal thread even if it is in the middle of running. The below scripts show the semantics.

asynchprinter = demo::AsynchPrinter();

for( i=1; i<11; ++i)

asynchprinter.obj = i;

And demo::AsynchPrinter is defined as below.

namespace demo;

asynch struct AsynchPrinter(input obj;)

as process

{

std::println(obj = input.obj);

}

It may surprise some user that the above script may print nothing, instead of a complete 1 to 10 integer list. The reason is that the main thread destroys the “asynchprinter” objet before it can finish its internal thread can finish its job. Modify the script to use the “waitfinish” member function of Asynchronous Structure will fix the problem.

asynchprinter = demo::AsynchPrinter();

for( i=0; i<10; ++i)

asynchprinter.obj = i;

asynchprinter.waitfinish();

The waitfinish member function will only return when the input queue of the structure is empty and there is at lease one thread inside sleeping and waiting for new input. For our example, it will guarantee that all objects from 1 to 10 are printed before the script quits.

7.4 The Life Cycle of Asynchronous Structure

We already know that you can trigger the evaluation of synchronous structure by setting its input property and/or by explicitly making structure call. And we expect the evaluation to finish before the execution flow goes further down.

Asynchronous Structure is different by nature since it has a live thread inside. The following describes the basic behavior of Asynchronous Structure.

When Asynchronous Structure is created, its internal thread is inactive.

x = demo::asynchinput(); // no thread started here

Setting the input property or getting its output property will trigger the internal thread to start running.

y = x.newinput; // this will trigger the internal thread running in x

You can manually force internal thread to start by calling the start() member function.

x.start(); // this will start the internal thread of x too

If an Asynchronous Structure has inputs, its internal thread always loops back to wait for new inputs automatically. It will stop otherwise.
When an Asynchronous Structure is out of scope, its destructor will be called and it will force the internal thread to stop, even if it is still active.
User can call the waitfinish() member function to wait for the internal thread of an Asynchronous Structure to become inactive. Inactive internal thread either sleeps waiting for input or completely stops.

x.waitfinish(); // this will block until internal thread of x becomes inactive

The assignment of one Asynchronous to a new variable will give the new variable a inactive structure with empty input and output queue. In another word, the input and output queue are copied.

y = x; // y will also be an asynch struct like, but no i/o queue and inactive

Asynchronous Structure does not support structure call. Making structure call to Asynchronous Structure will generate error value.

y = x( input1 = 0.0); // y will be of error value because x is aynch struct

7.5 Asynchronous Structure in Composition

Asynchronous structure and composition naturally work together. While in process script, user need to pay attention to the semantic difference of an asynchronous structure, in composition, user wire the asynchronous the same way as synchronous structure. Luban take care of the independent thread inside the asynchronous structure automatically.

In below example, we code one asynchronous structure to listen to network message. Then we use the structure in a composition.

namespace demo;

asynch struct messenger

(

output msg;

)

as process

{

socket = net::socket(“msgserver”, 6500);

while ( true )

{

obj = socket.readobj();

if ( obj == “good bye” ) break;

output.msg = obj;

}

struct listenNprint

()

as composition

{

MSG: ::messenger();

PRINTER: std::println(obj = MSG.msg );

}

Above code creates an asynchronous structure to listen to the port 6500 on server “msgserver”, and set the output “msg” whenever object arrives from the server. The “messenger” structure will only stops when it hears “good bye” from the server. The next composition uses this structure as its component and wire it together with a structure that prints message on console. We can run the composition structure using the below script:

demo::listenNprint(=);

When the composition “listenNprint” runs, it will listen to the message on network print what it hear on the console until the server says “good bye” to it.

7.6 Ad Hoc Asynchronous Cell In Composition

You can put ad hoc asynchronous cell into a composition by simple declare the ad hoc process cell as “asynch”. The code inside the cell will follow the asynchronous structure semantics automatically. We can code the “listenNprint” structure using ad hoc asynchronous cell as below:

namesapce demo;

struct listenNprint

()

as composition

{

asynch MSG: { socket = net::socket(“msgserver”, 6500);

while ( true )

{

obj = socket.readobj();

if ( obj == “good bye” ) break;

MSG = obj;

}

PRINTER: std::println(obj = MSG );

}

You can notice that in the ad hoc cell, there is no output property to set except itself. And its fellow component cells can only refer the value of itself instead of the output property.

7.7 Asynchronous Component and Cycle In Composition

As mentioned in previous chapter, cycle consists of only synchronous cells in a composition is not allowed. In another word, if the cycle has one or more asynchronous component inside, it is OK for Luban. Let’s look at the below example:

namesapce demo;

asynch struct player

(

input receive;

output send;

)

{

output.send = 1; // send a ball

while ( true )

{

ball = input.receive;

std::println(obj=ball);

if ( ball == 1000 ) break;

ball++;

output.send = ball;

}

struct pingpong

()

as composition

{

P1: ::player(receive = P2.send );

P2: ::player(receive = P1.send );

}

It is obvious that the composition structure “pingpong” has a cycle involving cell P1 and P2. Yet Luban is still going to happily take it and run through because P1 and P2 are asynchronous component, so it does not violate Luban’s no-cycle rule.

If you run “pingping” structure, it will print out number 1 to 1000, and print each number twice in a row.

8 Error Handling of Luban

The web of our life is of a mingled yarn, good and ill together.

- William Shakespeare

Luban use a special type: error type to handle errors. The value of error type is generated as error happens in Luban code evaluation. The basic principle of Luban error handling is very simple and can be described in below two points:

· Propagate error value along the line of expression, assignment and control flow if sensible.

· Abort the evaluation of Luban structure or script if the error can not be propagated in a sensible way.

8.1 So Long Try and Catch Madness

There are enough “who catch what at where” confusion about exception in many different languages. Luban goes a different way, it does not have exception at all, neither try and catch statements. The basic idea is that to use error value to represent error is simpler and easier to understand. Error value also fits Luban’s data flow model very well.

8.2 Error Value Generation and Propagation

// below variables all contain error

firsterr = 1/0;

alsoerr1 = firsterr++;

alsoerr2 = firsterr + alsoerr1;

test = firsterr isa error and alsoerr1 isa error and alsoerr2 isa error; // test = true

In above script, first error is generated by operation to divide with zero. Then the error propagates through a series of expression and assignment statements. So the error happens and spread logically corresponding to the code structure. Responsible user can decide to check error and take action at any point.

8.3 Errors That Stop The Show

There are certain points that is error sensitive in Luban. When errors happen at those points, Luban can only stop the evaluation of the current structure or script and report error. These points are listed as following.

Failure of assignment is one of the error sensitive points. Failure of assignment can be caused by assign object of wrong type to typed variable or structure property, for example:

int intvar;

intvar = 3.1416; // world ends here, assign double to a int type variable

std::println(obj=”Devil shall never see the light”); // never reach this line

If the result of condition expression inside control flow statement is not of bool type, Luban will stop the evaluation because it can not determine the control flow.

X = 0;

if ( X ) // X is not bool type, evaluation stop

X++;

When the call to a process structure stops because of error, the call returns error value instead of a fully evaluated structure.

Except the above error sensitive points, when error happens Luban will happily continue and pass the error object around. It is totally up to the user to decide where they want to check the error or if they want to check the error at all.

9 Utility Data Types and Structures

Men have become the tools of their tools.
- Henry David Thoreau

There are a set of commonly used data types and structures packaged together with Luban programming language. Though they are not part of the Luban programming language, they are so useful that we need to use a chapter to describe them.

9.1 Console, File and Socket

These are the basic data types used to persist information or pass them around network. And they share a more or less common interface. Below listed are their Luban type names and the description of their member functions.

Luban type names.

Console: std::console

File: std::file

Socket: net::socket

Member functions that apply to all above data types:

write(obj1,obj2…):

Writes the object in human readable string format to the media.

Return null if successful, or error type in case of failure.

writeline(obj1,obj2…):

Same as write, only adds one extra line break at the end of each object string.

Return null if successful, or error in case of failure.

read( int n=1 ):

Reads specified number of characters, or read one character if n not specified.

Returns a string containing all characters read, or error in case of failure.

readline():

Reads the media to break of line.

Returns a string, or error in case of failure.

writeobj(obj1,obj2…):

Writes the objects in platform independent binary format from which the object can be restored.

Returns null if successful, or error for failure.

readobj():

Restore object back from media, likely the object written by writeobj.

Returns restored object.

What need to be mentioned is that writeobj and readobj are designed to be used in pair. Any object serialized and persisted with writeobj can be restored using readobj.

9.2 How To Create, Read and Write File

Below is a simple text file processing example:

linecount = 0;

wordcount = 0;

txtfile = std::file(“mydatafile”, ‘r’); // open file ”mydatafile”

alldata = txtfile.readall(); // read ALL content of the file into alldata

lines = alldata.split(‘\n’); // split alldata into vector lines

foreach( line in lines ) // Iterate through each line in the vector lines

{

++linecount; //increase line count

words = line.split(); // Split one line into a vector of words

wordcount += words.size(); // increase word count

}

//print result

std::println(obj=”lines: “+string(linecount)+” words: “+string(wordcount));

It needs to be mentioned that above code use a new file type member function readall() that reads the whole content of a file into a string data object. The code then split the whole content into lines and does analysis.

The below example script shows how to use file to persist and restore data object using file:

fw = std::file(“objfile”, ‘w’); // open a file for writing

fw.writeobj([1,2,3]); // write a vector object

fw.close(); // close the file

fr = std::file(“objfile”, ‘r’); // open file for reading

vec = fr.readobj(); // read one object back

std::println(obj=vec); // print out [1,2,3]

There are three modes can be used to open a file. They are read, write and append, and are represented by characters as ‘r’, ‘w’ and ‘a’. Above code shows read and write. While in the write mode opening an existing file will cause it to be truncated, opening the same file in append mode will cause the later writing append to the end the file.

9.3 Socket Routine, Server and Client

The below scripts demonstrate how to use socket, including both client and server side. The server side uses a new data type net::listener that is basically a TCP port listener. It can be created given a port number. Its major member function is accept() that accepts one incoming connection and return a socket object that can talk to the client.

Server script:

// This server greets clients and assign each of them a number

listener = net::listener(6500); // take TCP port 6500

clientcount=0;

while( true ) // forever loop

{

socket = listener.accept(); // wait for client connection

name = socket.readobj(); // read client name

socket.writeobj(“Hello, ”+name); // say hello

++clientcount; // increase client number

Socket.writeobj(clientcount); // send the number to client

}

Client script:

socket = net::socket(“localhost”, 6500); // connect to server at port 6500

socket.writeobj(“Chinese”); // send server its name

msg = socket.readobj(); // receive greeting

std::console().writeline(“The server says: “, msg); // print greeting

mynumber = socket.readobj(); // read assigned number

std::console().writeline(“My number is “, mynumber); // print number

In the above, the server script takes TCP port 6500 and listens to incoming connection. For each client connection server says hello then assign a number to the client. Client script simply connect to the server at port 6500, send its name, then receive greeting message and number. You can run both scripts on the same machine. If you want to run them on different machines, you need to change the sever name from “localhost” to the actual server host name.

9.4 Utility Structures: std::println and std::print

Just for convenience, std::println and std::print structures are coded as part of Luban package. The structure print object to console in a simple way. Below is their source code:

namespace std;

struct println( input obj;)

as process

{ std::console().writeline(input.obj); }

struct print( input obj;)

as process

{ std::console().write(input.obj); }

9.5 Utility Structure std::des

Every Luban object can dump itself into a binary stream. In order to restore one object out of a binary string, the std::des structure is needed, as below sample code shows.

x = {“one”:1, “two”:2}; // construct a map

xguts = x.serialize();// serialize it into a string

xback = std::des(stream=xguts).obj; // call std::des to restore it

truth = xback == x; // truth = true, compare the restored object and original

10 Add New Data Type To Luban

The head learns new things,

but the heart forever practices old experiences.
- Henry Ward Beecher

This chapter demonstrates a way to code a C++ class and export it to Luban. This chapter is only for hard core professional who wishes to export C++ classes into Luban.

Basically the adding of new data type can be done in two steps. First is to code the class in a Luban compliant way, second is to let Luban know where to load the class.

10.1 Code a Luban Data Type In C++

Here the example is to code a new Luban data type demo::greeter. This data type does the simple job to print out “Hello, world” upon request.

To code any Luban compliant data type in C++, the convention is to always use two source files, one header file and one source module. In our simple greeter case, we will code header file greeter.hpp and greeter.cpp.

Below is the source code in greeter.hpp:

#include <lbtypes/lbobject.hpp>

#include <lbtypes/LBDeclareMacro.hpp>

class Greeter : public Luban::LBObject

{

LBDECLARE( Greeter )

public:

string toString() const; // for human readable printing

ostream& toStream(ostream& o) const; // serialize guts

istream& fromStream(istream& i, int major=-1, int minor=-1); // restore

virtual bool equals(const LBObject& another) const;

// member function visible in Luban

LBObject *luban_greet(const LBVarArgs *args);

};

What the header file tells us:

· All Luban data types are derived from Luban::LBObject data type.

· A mysterious macro “LBDECLARE()” needs to be put in.

· There are four functions must be implemented for all Luban data types.

“toString()” is used to convert the data object to a human readable string, and this function makes all Luban types printable and castable to string type.

“toStream()” is used to serialize the internal content of the data object. This function makes Luban types universally serializeable.

“fromStream()” is used to restore content from a serialized data stream which is produced by the “toStream()” function in the same class. Please note that “toStream()” only needs to be responsible to “fromStream()” function, meaning that the stream from “toStream()” can be used by “fromStream()” to restore to the original state.

“equal” function offers the comparison among data objects in the same class or different classes.

The conclusion, a Luban data type is required to know minimum three things:

1. Print itself

2. Serialize/de-serialize itself

3. Check equality against others

· The last member function “luban_greet” bears a signature that is required for any luban class member function to be exported to Luban language.

Then next step is to define these in a CPP module. Below is the source code in greeter.cpp:

#include “greeter.hpp”

#include <lbtypes/LBDefineMacro.hpp>

LBDEFINE_EXPORT(Greeter, “demo::greeter”, 1, 0 )

LBDEFAULT_STATIC_CONSTRUCTOR(Greeter);

string Greeter::toString() const

{

return “Greeter“;

}

ostream& Greeter::toStream(ostream& ost) const

{

return ost;

}

istream& Greeter::fromStream(istream& ist, int major, int minor)

{

return ist;

}

bool Greeter::equals(const LBObject& another) const

{

return dynamic_cast<const Greeter*>(&another) ;

}

LBEXPORT_MEMBER_FUNC(Greeter::luban_greet,“greet”, “string greet()”)

LBObject *Greeter::luban_greet(const LBVarArgs *args)

{

if ( args == 0 || args->size() == 0 )

return new LBString(“Hello, World”);

return new LBError(“demo::greeter() function takes no arguments”);

}

For the greeter.cpp module, the first thing needs to be put in is the LBDEFINE_EXPORT macro. From its appearance it indicates the mapping from C++ class name to Luban data type name plus the version number, and the version number can be used by the fromSteam() member function to decide what to do when restoring a data object stream of different version.

The line after the LBDEFINE macro is another macro LBDEFAULT_STATIC_CONSTRUCTOR that defines a default form of static constructor that does nothing more than calling the default constructor. More details of static constructor can be found in next section.

greeter.cpp also defines the mandatory virtual functions. toString() function returns a simple string to indicate itself as a greeter object. ToStream, fromStream functions do virtually nothing because there are no internal data to be serialized or restored. Equals() function check if the object compared against is a Greeter type, and returns true if it is.

The most interesting function here is the luban_greet. Luban requires that the member functions to be exported to Luban to have the same signature, which luban_greet bears. Basically the function signature can pass any number of arguments of any type, though the member function itself has to check the number and type of arguments then either perform requested action or return error value.

To export a member function to Luban, the mapping from the C++ member function name to the Luban member function name needs to be indiciated. That is what the LBEXPORT_MEMBER_FUNC macro does. The first argument is the C++ member function name, then corresponding name in Luban. The last argument is the synopsis of the function.

After the coding, we can compile our new class into a shared library. In our case we named the library libgreeter.so. That finishes all we need to do for C++ coding of a new Luban data type.

10.2 Inform Luban About The New Type

Next thing we need to do is to tell Luban how to load this new data type. For this step all we need to do is to list the new data type’s C++ class name, its Luban type name and the shared library name in a file, then tell Luban to read the file at starting. Luban has a dynamic data type loading mechanism that will load the shared library when the data type is needed at run time.

For our new demo::greeter type, we need to put the below line into a file.

Greeter libgreeter.so demo::greeter

The format of the line is C++ class name, shared library name then Luban type name. Let’s write the line into a file named “import_types”.

Now everything is ready, we can test to use this new type in Luban, like below lines show:

Mycomputer > luban –t import_types –i

Ø s

g = demo::greeter();

msg=g.greet();

Hello, World

>quit

MyComputer>

In the above example, Luban interpreter is started with –t flag that tells Luban to read a file of imported types. In our case there is only one data type demo::greeter in the file. At Luban interpreter command line, we use “s” command to start scripting. Then Luban runs the script and print out the result of the last evaluation that is the famous “Hello, World” message.

10.3 Basic Operations

For this new demo::greeter type, only a minimal basic operations can be performed. These operations include assignment, equality comparison, casting to string, serialize to stream, restore from stream and member function call. The below sample code shows the basic operations.

X = y; // assignment

X == y; // equality check

x.y(); // member function call

string(x); // cast to string

There is a set of standard member functions available for all Luban data types:

H = x.hash(); // hash key for the object, default value 0

Sz = x.size(); // apply to container and string type, default value error

x.contains(y); // apply to string and container type

x.clear(); // apply to string and container

guts = x.serialize(); // returns the binary guts stream in a string

h = x.invoke(“hash”); // call hash() in a reflective way

funmap = x.allfuncs(); // map containing all member functions and descriptions

obj = std::des(stream=guts).obj; // restore object from a string

10.4 More Operations

One obvious way to expand available operation is to add and export more member functions. There is also a set of virtual functions that can be implemented so more Luban expression operators can be applied on the data type. These functions are optional. If they are not implemented, the corresponding operation will result in error value. The following lists the virtual functions and their corresponding operation in Luban.

void plus(const LBObject& x) +, +=

void minus(const LBObject& x) -, -=

void mul(const LBObject& x) *, *=

void div(const LBObject& x) /, /=

void mod(const LBObject& x) %, %=

void neg(); -x

void plusplus(); ++

void minusminus(); --

bool before(const LBObject& x) <,>,<=,>=

//container related functions

int hash() //used as hash key when the object is used as key in a map

const LBObject* index(const LBObject* idx) // a=b[c];

LBObject* index(const LBObject* idx) // b[c] += 1;

bool replace(const LBObject* idx, const LBObject* value); // a[b]=c;

LBConstIterator *getIterator(); // foreach( x in vectory )

int size(); // number of elements

10.5 Static Constructor And Casting

As previously mentioned, Luban requires all data types to have a static constructor. Luban static constructor is simply the function Luban calls to construct the object upon request. Technically it is a relay between Luban and the real C++ constructors of the data type. There is a default implementation available in the format of a macro. But it does nothing more than calling the default constructor of the class.

You need to implement your own static constructor if you need to put arguments into object construction. A sample static constructor is like below:

// sample static constructor for a fake LBComplex type

LBObject *LBComplex::staticConstructor(const LBVarArgs *args)

{

int numarg = args? args->numArgs():0;

switch ( numargs )

{

case 0:

return new LBComplex();

case 1:

const LBDouble *dbl = dynamic_cast<const LBDouble*>(args->getArg(0));

if ( !dbl )

throw LBException(“Wrong type”);

return new LBComplex(double(*dbl));

case 2:

const LBDouble *dbl1 = dynamic_cast<const LBDouble*>(args->getArg(0));

const LBDouble *dbl2 = dynamic_cast<const LBDouble*>(args->getArg(1));

if ( !dbl1 || !dbl2 )

throw LBException(“Wrong type”);

return new LBComplex(double(*dbl1), double(*dbl2));

}

throw LBException(“Wrong number of args”);

}

With above static constructor coded, user can construct complex type in Luban like below:

C1 = math::complex(); // default

C2 = math::complex(1.0); // complex with only real part

C3 = math::complex(1.0, 2.0); // both real and imaginary

From above example, we can see static constructor also perform the task of type casting, in the case of single argument to the constructor. The variable C2 contains a complex cast from a simple real number.

If we wish to cast a complex to a double, using only static constructor, we have to change the static constructor of double type, which we may not be able to do because double is a built-in type and the code is out of the reach of us. Luban offers an alternative way to implement casting for the situation. That is to implement the below virtual function.

void LBComplex::cast(LBObject *castto)

{

LBDouble * dbl = dynamic_cast<LBDouble *>(castto);

if ( ! dbl )

throw LBException(“Wrong type to cast to”);

*dbl = LBDouble( _real );

}

Above function is to cast the real part of the complex into the double number passed in. It will enable the below construction in Luban:

C = math::complex(1., 2.);

D = double( C );

There could be the case that both the static constructor and cast function can be applied to the same operation. And the rule is to choose static constructor over cast function.

10.6 Helper Classes

There are two helper classes in Luban C++ importing mechanism. One is class Luban::LBVarArgs that is used majorly to pass data into member functions. Luban::LBVarArgs is an abstract class.

class LBVarArgs

{

public:

virtual int numArgs() const = 0;

virtual const LBObject* getArg(int index) const = 0;

};

Another helper class Luban::LBConstIterator is also an abstract class. It is used by container data type to return iterator for Luban’s foreach statement.

class LBConstIterator

{

public:

virtual bool next()=0; // move forward

virtual const LBVarArgs* getCurrentRow() = 0;

};

Luban::LBConstIterator views each container as a number of rows and each row can contain one or more elements. It uses LBVarArgs interface to access elements in a row. The need to implement your own LBConstIterator derived class may arise when your implement your own specific container class and export to Luban.

10.7 Common Assumptions About Luban Data Types

Luban takes certain common assumptions about the data types imported from C++. The assumptions are actually in line with C++ common requirements for classes. They are listed as below.

· A good copy constructor

This is mandatory. All types must have a working copy constructor.

· An assignment operator

This operator will be called. Though you may implement one that does nothing more than throwing exception if necessary. Luban uses this operator to improve efficiency of variable assignment. But everything still works if the assignment operator fails.

· A good default constructor

This is necessary when you use the default static constructor, or your own static constructor calls the default constructor explicitly. Also it is needed for streaming.

· A good destructor

· No worry about multi-threading ( usually )

Though Luban is a multi-threading language, it has a mechanism to ensure that only one thread calls member function of an object at any time. So the coder of the class normally does not need to worry about thread safety, unless you use some object sharing technology like reference counting. If you do sharing object among different instance of your class, then you need to do something make sure your class is thread safe. Because there will be more than one thread accessing your shared object.

11 Reflection

Now he worships at an altar of a stagnant pool

And when he sees his reflection, he's fulfilled.

- License to Kill, Bob Dylan

A complete dynamic reflection mechanism has been built into Luban language. All objects and components can be dynamically constructed by string name that is only determined at run time. All structure properties can be dynamically set and get with string property name. And all object member functions can be called via string name too. The below sections describe more details.

11.1 Construct Object Using Reflection

You can construct any data objects and components by their full names. The proxy for reflective object and component construction is a data type named “luban::ns” that represents the namespace of Luban. By calling the create() member function of luban::ns object with the name string of object and component type, you can construct objects and components dynamically, as the below sample Luban script shows.

// make a namespace object

ns = luban::ns();

//Basic type construction

pi = ns.create("double", 3.1415);

i = ns.create("int", 3);

b = ns.create("bool");

c = ns.create("char", 'a');

s = ns.create("string", "Hello, Reflection")

v = ns.create("vector", 10, 0);

m = ns.create("map");

s = ns.create("set");

socket = ns.create(“net::socket”, “localhost”, 8500);

//component construction

adder = ns.create("luban::demo::simpleadder");

adder.(op1=100, op2=200);

std::println(obj="component output: "+string(adder.result));

From above code you can see you can call the object construction by its type name and pass in arbitrary number of arguments following the string type name. The whole mechanism is dynamic. The below code constructs a default instance of the same type as variable x whatever the type is.

xdefault = luban::ns().create(string(typeof(x));

11.2 Get Type Information Object Using Reflection

Not only you can create any object instance using reflection, you can also get the type information object using reflection and use that object for type checking purpose. Take a look below code.

ns = luban::ns();

filetype = ns.gettype(“std::file”);

fileobj = std::file(“xfile”, ‘r’);

truth = fileobj isa filetype; // truth = true

In above code, a “filetype” object is created by calling “gettype” member function of luban::ns. The created object is then used to check the type of a variable named “fileobj”.

11.3 Set and Get Structure Property by Reflection

The component model of Luban is based on its struct type that is property value flow based. The Luban’s reflection mechanism covered set and get of struct properties by reflection, as the below code shows.

adder =luban::demo::simpleadder();

adder.pset("op1", 1.0);

result = adder.pget("result");

std::println(obj="component output: "+string(result));

adder.pset({"op1":1.0, "op2":2.0});

result = adder.pget("result");

std::println(obj="component output: "+string(result));

adder.pset(["op1",”op2], [1,2]);

result = adder.pget("result");

std::println(obj="component output: "+string(result));

inputs = adder.pget(["op1","op2"]);

std::println(obj="component inputs: "+string(inputs));

Above code shows that you can set and get struct properties by using its built-in member function pset and pget, with property name string passed in as argument. You can set multiple properties at once by passing a map with property names as keys or a vector of keys plus a vector of values. To get property values pget function can be used and if multiple values need to be retrieved at once a vector of property names can be used as arguments. And a vector of values is returned as result.

11.4 Call Object Member Function by Reflection

The member function of any Luban data type can be called via reflection too. As below code shows.

mp = {"John":1, "Paul":2, "George":3};

mp.invoke("insert", "Yoko", 4);

std::println(obj=mp);

The special member function invoke is the entrance for relective member function call, and invoke is available for all Luban types. In the above example, the insert member function of map type is called through invoke and a pair of key and value is inserted as result.

There is one related member function allfuncs() that is available for all types in Luban, which is to give out a complete map of all available member functions and their description in a map. Below code shows how to use the special allfuncs().

funcmap = anyobj.allfuncs();

11.5 Reflection Usage Example: Remote Component Call

Using Luban’s reflection mechanism and network streaming facilities, it is almost trivial to construct a client/server tool to enable remote calling of Luban components. On the client side, all need to be done is to send the component name, input property values plus the desired output property names. For the server, it needs to listen to the incoming connection get the component name, inputs and desired output names. Server then builds the component by name, set its inputs, get its output as required and send results back through the same socket. The whole system can be built with very little Luban code. That’s why we can list the whole system code as in less than two pages below.

11.5.1 Remote Component Call Server Code

// a multi-threaded RCC server

namespace luban::demo;

struct RCCServer( store readwrite int port;)

as process

{

rootlistener = net::listener(store.port);

while ( true )

{

onesock = rootlistener.accept(); // receive incoming call

::HandleRCC(socket=onesock) & // dispatch to a different thread

}

// HandleRCC actually handle one call and return results

struct HandleRCC

(

input:

socket;

)

as process

{

socket = input.socket;

package = socket.readobj(); // get the whole package

compname = package[0]; // first one is component name

args = package[1]; // second one input properties

resultprps = package[2] //third one for desired outputs;

comp = luban::ns().create(compname); // construct by reflection

comp.pset(args); // property setting by reflection

result = comp.pget(resultprps); // property get by reflection

socket.writeobj(result);

}

There are two struct are defined in above code. One is the main loop of the RCC server, all it does is to listen to the incoming connection and dispatch to the struct HandleRCC in a different thread then back to listening again. Thread dispatching is a common technique used in server coding to accommodate more than one request simultaneously.

In the struct HandleRCC, a package is retrieved from socket. And component name, inputs and desired outputs are unpacked. Then using reflection mechanism, component is built, its inputs are set and outputs are gotten. Then the results are written back to the same socket.

11.5.2 Remote Component Call Client Code

// Remote Component Call client code

namespace luban::demo;

struct RCCClient

(

store readwrite int port; store readwrite string host;

input:

string compname;

map ins;

outs;

output results;

)

as process

{

onesock = net::socket(store.host, store.port);

onesock.writeobj( [input.compname, input.ins, input.outs ] );

output.results = onesock.readobj();

}

The RCC client code is even simpler. All it does it to make a connection to the specified server host and port. Then client put user specified component name, inputs and desired outputs into a package of vector format, write to the socket. The last step is simply waiting for the results to come back from server.

11.5.3 How to Run RCC Server and Client

The below line of Luban script will start the RCC server at port 9600.

luban::demo::RCCServer(port=9600);

The below several lines of code dispatch the work separated to multiple parts and send them simultaneously to the server and wait for them to finish.

client = luban::demo::RCCClient();

client.host = “superserver.mycompany.com”;

client.port = 9600;

P1=client(compname=”myns::supercalc” , inputs={“vectosum”:[1,2,3,4]}, outs=”sumofvec”).results &

P2= client(compname=”myns::supercalc”, inputs={“vectosum”:[5,6,7,8]}, outs=”sumofvec”).results &

P3 = client(compname=”myns::supercalc”, inputs={“vectosum”:[9,10,11,12]}, outs=”sumofvec”).results &

waitfor P1,P2,P3;

total = P1+P2+P3;

Above code construct a RCC client to talk the server running on “superserver.mycompany.com” at port 9600. Through the connection, client calls the component named “myns::supercalc” to summarize numbers in a vector. It dispatches three calls in different thread to summarize three vectors simultaneously, wait for them to come back. Finally it then adds up the three results to get the grand total.

11.5.4 Possible Improvements of Sample RCC

Using reflection and networking, we build a very practical and useful RCC client/server system. There many ways to improve it in practice. One straight forward way is to run the RCC server on a virtual server that may dispatch jobs to a server cluster. Since cluster can probably handle the load balancing by itself, it is transparent to Luban user. For Luban user the unlimited power of parallel computer is now available at their finger tips.

11.6 Reflection and Parallel Computing: A Perfect Match

From above example RCC system, we can see that it is powerful to combine reflection and parallel network computing together in Luban. Luban makes a perfect environment to dispatch parallel jobs and synchronize them. With Luban, you have the infinite capacity of cheap computing power from server farm. You can attach problem of any complexity as far as the problem itself can be divided and conquered as smaller problems.

12 Luban Interpreter

Simplicity in character, in manners, in style;

in all things the supreme excellence is simplicity.
- Henry Wadsworth Longfellow

To run Luban code, we use a simple Luban interpreter named, surprise, “luban”. This luban interpreter takes Luban code in the files and runs them through.

The same Luban interpreter can also run in interactive mode. In this mode you can explore the content of Luban name space, and do interactive scripting. The things the interpreter can do are actually quite simple. Though I believe it serves its purpose.

Below are the command line arguments that can be used to control the behavior of interpreter.

· -t <filename>

This flag is to specify the name of the file that lists all the data types imported from other languages like C++ in the form of shared library. If this argument is not specified, Luban uses a default imported type list file, which is either /usr/lib/luban/imports or environemtn variable $LUBAN_HOME/imports. More details of data type importing can be found in chapter 10.

· -i

When this flag is set, Luban interpreter will open a interactive command line to allow interactive scripting and name space exploring, like the below example shows:

Mycomputer > luban –i

> list std

std::file std::file

std::console std::console

> script

std::console().writeline(“Hello, world”);

ESC\ENTER

Hello, world

> quit

Mycomputer >

The above example starts Luban interpreter with interactive command line. At the Luban command line, user then use “list” or “l” command to list the components with “std” name space, and “script” or “s” command to type in and run lines of Luban code. Press ESC key then ENTER to get out of scripting mode and back to interactive command prompt. At the end user uses the “quit” command to terminate the session and back to operating system. So in this interactive mode, you can checkout the content of name space and you can script and that’s about it.

For Luban Beta 1.2 and later version, there is a new command “e” or “edit” to invoke an external editor to edit Luban script and run it. The editor can be specified by user with EDITOR environment variable. Luban will try to use “vi” if EDITOR environment variable is not specified. This feature works on Linux and Windows with Cygwin installed. It will not work for Luban Windows binary package without Cygwin installation.

· -s <Luban code>

Use this flag you can put a short line of Luban code in the command line when starting Luban interpreter. And Luban interpreter will run the code, like below example:

Mycomputer > luban –s “std::console().writeline(100);”

100

Mycomputer >

· All other arguments other than above flags will be taken as Luban source code file names. And all the non-component Luban script files will be run in the order of their command line position.

13 Sample Application: Topic Based Messaging System

The power of role model is infinite.

- Mao Zedong

Using Luban programming language and its networking and file utilities, it is very easy to write a simple topic based messaging system. Client can subscribe and publish message by topic, while server accepts client subscription, receive topic message update and broadcast to subscribing clients. The code required to implement such a messaging system is very small. And they are completely listed below.

13.1 Messaging Server Code

namespace luban::demo;

stationary struct MsgServiceConfig

(

static readonly host = "localhost";

static readonly subport = 9876;

static readonly pubport = 9877;

static topics = {:};

);

struct SubService( )

as process

{

subport = net::listener( ::MsgServiceConfig.subport );

while ( true )

{

sck = subport.accept();

topic = sck.readobj();

std::println(obj=”subscriber for “+topic);

first = false;

if ( not ::MsgServiceConfig.topics.contains( topic ) )

{

::MsgServiceConfig.topics[topic] = [null, []];

first = true;

}

::MsgServiceConfig.topics[topic][1].append(sck);

sck.writeobj("OK");

if ( not first )

sck.writeobj( ::MsgServiceConfig.topics[topic][0] );

}

struct PubService( )

as process

{

pubport = net::listener( ::MsgServiceConfig.pubport );

while ( true )

{

sck = pubport.accept();

pack = sck.readobj();

std::println(obj=pack);

topic = pack[0];

value = pack[1];

sck.writeobj("OK");

if ( not ::MsgServiceConfig.topics.contains(topic) )

{

::MsgServiceConfig.topics[topic] = [ value, [] ];

continue;

}

::MsgServiceConfig.topics[topic][0] = value;

foreach( sub in ::MsgServiceConfig.topics[topic][1] )

sub.writeobj(value);

}

struct PubSubServer()

as process

{

::PubService(=) &

::SubService(=) &

}

The messaging server code consists of several components. The first one MsgServiceConfig is actually a shared data holder. It contains several static properties that are accessible for cross different Luban structures. The static properties save the information like server host name and port number. There are two ports open on the server, one to accept incoming subscription, another one to accept incoming topic updates. The essential data property is the one named “topics”. It is initialized as an empty map, it is to contain all the topics and their value, plus all the subscriber’s socket data. The key to the map is message topic. And the value of in the map is a two element vector. The first element is the value of the topic, while the second one is a vector containing all the subscriber’s connection sockets.

The execution logic of the server is coded into two structures. One structure “SubService” is to handle subscription request. It accepts the incoming request, look up the topic in the shared data map and put the socket into the vector associated with the topic subscribed. Another structure “PubService” accepts the message topic updating request. It takes the topic and value of the update, looks up the data map, update the value of the topic, then send out the updated message to each of the subscriber through the saved socket vector.

13.2 Messaging Service Client Code

namespace luban::demo;

asynch struct Subscriber

(

input:

string topic;

output:

updates;

)

as process

{

socket=net::socket(::MsgServiceConfig.host,::MsgServiceConfig.subport );

socket.writeobj(input.topic);

ack = socket.readobj();

std::println(obj=ack);

while ( true )

{

newdata = socket.readobj();

output.updates = newdata;

}

struct Publisher

(

input:

string topic;

updates;

output:

serverack;

)

as process

{

socket=net::socket(::MsgServiceConfig.host,::MsgServiceConfig.pubport );

socket.writeobj([input.topic, input.updates]);

ack = socket.readobj();

std::println(obj=ack);

output.serverack = ack;

}

struct SubClient

(

input:

string topic;

)

as composition

{

sub: ::Subscriber(topic=input.topic);

printer: std::println(obj=sub.updates);

}

struct PubClient()

as process

{

for(;;)

{

console = std::console();

line = console.readline();

tandv = line.split();

ack = ::Publisher(topic=tandv[0], updates=tandv[1]).serverack;

console.writeline(ack);

}

The above client code are separated into “Subscriber” and “Publisher”. The logic of both structure are very simple. Subscriber connects to the message server on specific host and port, send the topic to subscribe to, then it holds the socket and listen to any update and send received update its output property so the downstream components can process the update. You may notice that this “Subscriber” structure is asynch. And it is purposely design to be so. Asynch structure fit naturally with network messaging service because it runs in a different thread and put all its output in queue. The “Publisher” also connects to the message server though on different port. It simply sends the topic of message and the message itself in a two element vector. “Publisher” is a regular structure, instead of asynch.

The other two structures are for testing purpose. The “SubClient” structure is compositional, it simply connects a “Subscriber” to a “std::println” structure. So every updates the “Subscriber” gets from network will be printed out to standard output by the “std::println” structure. The “PubClient” simply read line from standard input in the format of “topic message” and call the “Publisher” structure to publish the topic and message.

13.3 Potential Improvements to Messaging System

You can see from above sample code that the messaging system is practically useful though the coding of it is very simple. This simple layer present the network communication function in a topic based format that is far more simpler and easier than the socket.

One potential improvement to this simple messaging service is to expand it to a messaging/database hybrid system. Put it simply, if we persist all the message to a file system, it becomes a simple name key based database system. And things saved in it can be any Luban object. The hybrid system can be very interesting to use because it not only saves the data by name, it can broadcast the updated new data to subscribers cross the network.

14 Todos

I got a head full of ideas,

And that are drivin’ me insane.

- Maggie’s Farm, Bob Dylan

There are many things that can be done to fully realize or expand the power of Luban programming language and paradigm. A few of them are listed below.

· Luban Studio: A spreadsheet like IDE

This is an integrated part of Luban development environment. It is actually its IDE (Integrated Development Environment). The goal of this component is to hit home the intuitiveness and simplicity of Luban by presenting it in a GUI interface. One major advantage of this IDE is to enable Luban user to do composition in a way almost the same as spreadsheet construction. It will be immediately understandable for anyone who can use spreadsheet, even if he/she has never done programming before.

This Luban Studio may also include functionalities that are commonly used for programmers, including debugger, version control and cross network common component repository. The goal is to make Luban Studio a platform on which people build and share software components.

· GUI widget components for Windows, Linux and Java

It would be instantly gratifying for Luban programmers if they could easily construct an application with GUI within Luban’s infrastructure. Luban has built-in space designed to put GUI widget, which is asynchronous structure. The goal of this component is to wrap commonly used GUI widgets as asynchronous structures and put in Luban name space, including Windows, Linux and even Java GUI widget.

As soon as the GUI widget available as Luban components, they can be easily used as any other Luban components. Imagine wiring GUI components in a composition model, it is hard tot get anything more intuitive than that.

· Closure for Luban

It could be convenient to put a piece of code into an ad hoc closure with a name and run it as many times as you wish.

· Luban Java Bridge

It could be easier than Luban/C++ bridge because Java has a built-in reflection mechanism. This bridge will enable exporting Java classes and member functions into Luban, possibly including an API to let Java call Luban component too.

· Luban SQL Utility

It is simple to build a db.mysql data type for Luban to run SQL statement on a MySQL or any relational database server.

· wchar and wstring type