####1. What is K8?
K8 is a Javascript shell based on Google's V8 Javascript engine. It adds the support of flexible byte arrays and file I/O. K8 is implemented in one C++ source file. The only dependency is zlib in addition to V8.
####2. There are many Javascript shells with much richer features. What makes K8 special?
To some extent, Node.js, Narwhal, SilkJS, TeaJS and Sorrow.js are all Javascript shells. They not only provide binary storage and file I/O, the features available in K8, but also implement much richer functionality such as network I/O and database binding. However, most of the existing Javascript shells are designed for server-side applications, but not for general use cases as we do with Perl/Ruby/Python. Take the popular Node.js as an example. Node.js mixes file I/O and file system operations, two distinct concepts, in one File System module. In the module, we can either read an entire file or a fixed-length data blob, but are unable to read a line as is provided by most other programming languages. Many other JS shell implementations follow the CommonJS APIs, which have a similar problem: no usable APIs for general-purpose file I/O. After all these efforts, on file I/O, we even do not have a JS shell matching the usability of C, let alone high-level programming languages such as Perl and Python.
K8 aims to provide C-like file I/O APIs. It adds a File object for buffered
file reading and a Bytes object for flexible binary storage.
####3. How to compile K8? Are there compiled binaries?
You need to first compile V8 and then compile and link K8. Here is the full procedure:
git clone https://github.com/v8/v8 # download V8
(cd v8; git checkout 3.16.4) # k8 only works with 3.16.x
(cd v8; make dependencies; make x64.release) # compile V8
g++ -O2 -Wall -o k8 -Iv8/include k8.cc -lpthread -lz v8/out/*/libv8_{base,snapshot}.a
The two libv8*.a files should always be placed somewhere in v8/out, but
maybe in a deeper directory, depending on the OS.
Alternatively, you may download the compiled binaries for Mac and Linux from SourceForge. The source code is also included.
####4. An earlier version of K8 implemented a generic buffered stream. Why has it been removed?
To implement a generic buffered stream, we need to call a Javascript read
function in C++ and transform between Javascript and C++ data representation.
This procedure adds significant overhead. For the best performance on file
I/O, all the iStream functionality has been moved to File. Anyway, it
is not hard to implement buffered stream purely in Javascript.
All the following objects manage some memory outside the V8 garbage collector.
It is important to call the close() or the destroy() methods to deallocate
the memory to avoid memory leaks.
###Example
var x = new Bytes(), y = new Bytes();
x.set('foo'); x.set([0x20,0x20]); x.set('bar'); x.set('F', 0); x[3]=0x2c;
print(x.toString()) // output: 'Foo, bar'
y.set('BAR'); x.set(y, 5)
print(x) // output: 'Foo, BAR'
x.destroy(); y.destroy()
if (arguments.length) { // read and print file
var x = new Bytes(), s = new File(arguments[0]);
while (s.readline(x) >= 0) print(x)
s.close(); x.destroy();
}
###The Bytes Object
Bytes provides a byte array. It has the following methods:
// Create an array of type $type in length $len. $type can be: int8_t, uint8_t, int16_t,
// uint16_t, int32_t, uint32_t, float or double.
new Bytes(len, type)
// Equivalent to 'new Bytes(len, "uint8_t")'
new Bytes(len)
// Equivalent to 'new Bytes(0, "uint8_t")'
new Bytes()
// Property: get/set length of the array
.length
// Property: get/set the max capacity of the array
.capacity
// The index operator. If $pos goes beyond .length, undefined will be returned.
int obj[pos]
// Change the array type to $type, equivalent to changing the pointer type. .length and
// .capacity may be changed if the size of element is changed.
Bytes.prototype.cast(type)
// Equivalent to 'Bytes.prototype.cast("uint8_t")'
Bytes.prototype.cast()
// Deallocate the array. This is necessary as the memory is not managed by the V8 GC.
Bytes.prototype.destroy()
// Replace the byte array starting from $offset to $data, where $data can be a number,
// a string, an array or Bytes. The size of the array is modified if the new array
// is larger. Return the number of modified bytes. If only one byte needs to be
// changed, using the [] operator gives better performance.
int Bytes.prototype.set(data, offset)
// Append $data to the byte array
int Bytes.prototype.set(data)
// Convert the byte array to string
Bytes.prototype.toString()
###The File Object
File provides buffered file I/O. It has the following methods:
// Open $fileName under $mode. $mode is in the same syntax as fopen(). Integer $fileName for
// a file descriptor. In particular, 0 for STDIN, 1 for STDOUT and 2 for STDERR.
new File(fileName, mode)
// Equivalent to 'new File(fileName, "r")'
new File(fileName)
// Equivalent to 'new File(0)'
new File()
// Read a byte. Return -1 if reaching end-of-file
int File.prototype.read()
// Read maximum $len bytes of data to $buf, starting from $offset. Return the number of
// bytes read to $buf. The size of $buf is unchanged unless it is smaller than $offset+$len.
int File.prototype.read(buf, offset, len)
// Write $data, which can be a string or Bytes(). Return the number of written bytes.
// This method replies on C's fwrite() for buffering.
int File.prototype.write(data)
// Read a line to $bytes starting from $offset, using $sep as the separator. $sep==0 sets
// the separator to isspace(), $sep==1 to (isspace() && !' ') and $sep==2 to newline. If
// $sep is a string, the first character in the string is the separator. Return the line
// length or -1 if reaching end-of-file.
int File.prototype.readline(bytes, sep, offset)
// Equivalent to 'File.prototype.readline(bytes, sep, 0)'
int File.prototype.readline(bytes, sep)
// Equivalent to 'File.prototype.readline(bytes, 2, 0)'
int File.prototype.readline(bytes)
// Close the file
File.prototype.close()