Table of Contents

Introduction

This is the README file for the JavaScript Reference (JSRef) implementation. It consists of build conventions and instructions, source code conventions, a design walk-through, and a brief file-by-file description of the source.

JSRef builds a library or DLL containing the JavaScript runtime (compiler, interpreter, decompiler, garbage collector, atom manager, standard classes). It then compiles a small "shell" program and links that with the library to make an interpreter that can be used interactively and with test .js files to run scripts.  The code has no dependencies on the Navigator code.

Quick start tip: skip to "Using the JS API" below, build the js shell, and play with the object named "it" (start by setting 'it.noisy = true').

Build conventions (standalone JS engine and shell)

These build directions refer only to building the standalone JavaScript engine and shell.  To build within the browser, refer to the build directions on the mozilla.org website.

By default, all platforms build a version of the JS engine that is not threadsafe.  If you require thread-safety, you must also populate the mozilla/dist directory with NSPR headers and libraries.  (NSPR implements a portable threading library, among other things.  The source is downloadable via CVS from mozilla/nsprpub.)  Next, you must define JS_THREADSAFE when building the JS engine, either on the command-line (gmake/nmake) or in a universal header file.

Windows

Macintosh

Unix

Debugging notes

Naming and coding conventions

Using the JS API

Starting up

    /*
     * Tune this to avoid wasting space for shallow stacks, while saving on
     * malloc overhead/fragmentation for deep or highly-variable stacks.
     */
    #define STACK_CHUNK_SIZE    8192

    JSRuntime *rt;
    JSContext *cx;

    /* You need a runtime and one or more contexts to do anything with JS. */
    rt = JS_Init(1000000L);
    if (!rt)
        fail("can't create JavaScript runtime");
    cx = JS_NewContext(rt, STACK_CHUNK_SIZE);
    if (!cx)
        fail("can't create JavaScript context");

    /*
     * The context definitely wants a global object, in order to have standard
     * classes and functions like Date and parseInt.  See below for details on
     * JS_NewObject.
     */
    JSObject *globalObj;

    globalObj = JS_NewObject(cx, &my_global_class, 0, 0);
    JS_InitStandardClasses(cx, globalObj);

Defining objects and properties

    /* Statically initialize a class to make "one-off" objects. */
    JSClass my_class = {
        "MyClass",

        /* All of these can be replaced with the corresponding JS_*Stub
           function pointers. */
        my_addProperty, my_delProperty, my_getProperty, my_setProperty,
        my_enumerate,   my_resolve,     my_convert,     my_finalize
    };

    JSObject *obj;

    /*
     * Define an object named in the global scope that can be enumerated by
     * for/in loops.  The parent object is passed as the second argument, as
     * with all other API calls that take an object/name pair.  The prototype
     * passed in is null, so the default object prototype will be used.
     */
    obj = JS_DefineObject(cx, globalObj, "myObject", &my_class, 0,
                          JSPROP_ENUMERATE);

    /*
     * Define a bunch of properties with a JSPropertySpec array statically
     * initialized and terminated with a null-name entry.  Besides its name,
     * each property has a "tiny" identifier (MY_COLOR, e.g.) that can be used
     * in switch statements (in a common my_getProperty function, for example).
     */
    enum my_tinyid {
        MY_COLOR, MY_HEIGHT, MY_WIDTH, MY_FUNNY, MY_ARRAY, MY_RDONLY
    };

    static JSPropertySpec my_props[] = {
        {"color",       MY_COLOR,       JSPROP_ENUMERATE},
        {"height",      MY_HEIGHT,      JSPROP_ENUMERATE},
        {"width",       MY_WIDTH,       JSPROP_ENUMERATE},
        {"funny",       MY_FUNNY,       JSPROP_ENUMERATE},
        {"array",       MY_ARRAY,       JSPROP_ENUMERATE},
        {"rdonly",      MY_RDONLY,      JSPROP_READONLY},
        {0}
    };

    JS_DefineProperties(cx, obj, my_props);

    /*
     * Given the above definitions and call to JS_DefineProperties, obj will
     * need this sort of "getter" method in its class (my_class, above).  See
     * the example for the "It" class in js.c.
     */
    static JSBool
    my_getProperty(JSContext *cx, JSObject *obj, jsval id, jsval *vp)
    {
        if (JSVAL_IS_INT(id)) {
            switch (JSVAL_TO_INT(id)) {
              case MY_COLOR:  *vp = . . .; break;
              case MY_HEIGHT: *vp = . . .; break;
              case MY_WIDTH:  *vp = . . .; break;
              case MY_FUNNY:  *vp = . . .; break;
              case MY_ARRAY:  *vp = . . .; break;
              case MY_RDONLY: *vp = . . .; break;
            }
        }
        return JS_TRUE;
    }

Defining functions

    /* Define a bunch of native functions first: */
    static JSBool
    my_abs(JSContext *cx, JSObject *obj, uintN argc, jsval *argv, jsval *rval)
    {
        jsdouble x, z;

        if (!JS_ValueToNumber(cx, argv[0], &x))
            return JS_FALSE;
        z = (x < 0) ? -x : x;
        return JS_NewDoubleValue(cx, z, rval);
    }

    . . .

    /*
     * Use a JSFunctionSpec array terminated with a null name to define a
     * bunch of native functions.
     */
    static JSFunctionSpec my_functions[] = {
    /*    name          native          nargs    */
        {"abs",         my_abs,         1},
        {"acos",        my_acos,        1},
        {"asin",        my_asin,        1},
        . . .
        {0}
    };

    /*
     * Pass a particular object to define methods for it alone.  If you pass
     * a prototype object, the methods will apply to all instances past and
     * future of the prototype's class (see below for classes).
     */
    JS_DefineFunctions(cx, globalObj, my_functions);

Defining classes

    /*
     * This pulls together the above API elements by defining a constructor
     * function, a prototype object, and properties of the prototype and of
     * the constructor, all with one API call.
     *
     * Initialize a class by defining its constructor function, prototype, and
     * per-instance and per-class properties.  The latter are called "static"
     * below by analogy to Java.  They are defined in the constructor object's
     * scope, so that 'MyClass.myStaticProp' works along with 'new MyClass()'.
     *
     * JS_InitClass takes a lot of arguments, but you can pass null for any of
     * the last four if there are no such properties or methods.
     *
     * Note that you do not need to call JS_InitClass to make a new instance of
     * that class -- otherwise there would be a chicken-and-egg problem making
     * the global object -- but you should call JS_InitClass if you require a
     * constructor function for script authors to call via new, and/or a class
     * prototype object ('MyClass.prototype') for authors to extend with new
     * properties at run-time.
     */
    protoObj = JS_InitClass(cx, globalObj, NULL, &my_class,

                            /* native constructor function and min arg count */
                            MyClass, 0,

                            /* prototype object properties and methods -- these
                               will be "inherited" by all instances through
                               delegation up the instance's prototype link. */
                            my_props, my_methods,

                            /* class constructor properties and methods */
                            my_static_props, my_static_methods);

Running scripts

    /* These should indicate source location for diagnostics. */
    char *filename;
    uintN lineno;

    /*
     * The return value comes back here -- if it could be a GC thing, you must
     * add it to the GC's "root set" with JS_AddRoot(cx, &thing) where thing
     * is a JSString *, JSObject *, or jsdouble *, and remove the root before
     * rval goes out of scope, or when rval is no longer needed.
     */
    jsval rval;
    JSBool ok;

    /*
     * Some example source in a C string.  Larger, non-null-terminated buffers
     * can be used, if you pass the buffer length to JS_EvaluateScript.
     */
    char *source = "x * f(y)";

    ok = JS_EvaluateScript(cx, globalObj, source, strlen(source),
                           filename, lineno, &rval);

    if (ok) {
        /* Should get a number back from the example source. */
        jsdouble d;

        ok = JS_ValueToNumber(cx, rval, &d);
        . . .
    }

Calling functions

    /* Call a global function named "foo" that takes no arguments. */
    ok = JS_CallFunctionName(cx, globalObj, "foo", 0, 0, &rval);

    jsval argv[2];

    /* Call a function in obj's scope named "method", passing two arguments. */
    argv[0] = . . .;
    argv[1] = . . .;
    ok = JS_CallFunctionName(cx, obj, "method", 2, argv, &rval);

Shutting down

    /* For each context you've created: */
    JS_DestroyContext(cx);

    /* And finally: */
    JS_Finish(rt);

Debugging API

See the trap, untrap, watch, unwatch, line2pc, and pc2line commands in js.c. Also the (scant) comments in jsdbgapi.h.

Design walk-through

This section must be brief for now -- it could easily turn into a book.

JS "JavaScript Proper"

JS modules declare and implement the JavaScript compiler, interpreter, decompiler, GC and atom manager, and standard classes.

JavaScript uses untyped bytecode and runtime type tagging of data values. The jsval type is a signed machine word that contains either a signed integer value (if the low bit is set), or a type-tagged pointer or boolean value (if the low bit is clear). Tagged pointers all refer to 8-byte-aligned things in the GC heap.

Objects consist of a possibly shared structural description, called the map or scope; and unshared property values in a vector, called the slots. Object properties are associated with nonnegative integers stored in jsval's, or with atoms (unique string descriptors) if named by an identifier or a non-integral index expression.

Scripts contain bytecode, source annotations, and a pool of string, number, and identifier literals. Functions are objects that extend scripts or native functions with formal parameters, a literal syntax, and a distinct primitive type ("function").

The compiler consists of a recursive-descent parser and a random-logic rather than table-driven lexical scanner. Semantic and lexical feedback are used to disambiguate hard cases such as missing semicolons, assignable expressions ("lvalues" in C parlance), etc. The parser generates bytecode as it parses, using fixup lists for downward branches and code buffering and rewriting for exceptional cases such as for loops. It attempts no error recovery. The interpreter executes the bytecode of top-level scripts, and calls itself indirectly to interpret function bodies (which are also scripts). All state associated with an interpreter instance is passed through formal parameters to the interpreter entry point; most implicit state is collected in a type named JSContext. Therefore, all API and almost all other functions in JSRef take a JSContext pointer as their first argument.

The decompiler translates postfix bytecode into infix source by consulting a separate byte-sized code, called source notes, to disambiguate bytecodes that result from more than one grammatical production.

The GC is a mark-and-sweep, non-conservative (perfect) collector. It can allocate only fixed-sized things -- the current size is two machine words. It is used to hold JS object and string descriptors (but not property lists or string bytes), and double-precision floating point numbers. It runs automatically only when maxbytes (as passed to JS_Init()) bytes of GC things have been allocated and another thing-allocation request is made. JS API users should call JS_GC() or JS_MaybeGC() between script executions or from the branch callback, as often as necessary.

An important point about the GC's "perfection": you must add roots for new objects created by your native methods if you store references to them into a non-JS structure in the malloc heap or in static data. Also, if you make a new object in a native method, but do not store it through the rval result parameter (see math_abs in the "Using the JS API" section above) so that it is in a known root, the object is guaranteed to survive only until another new object is created. Either lock the first new object when making two in a row, or store it in a root you've added, or store it via rval.

The atom manager consists of a hash table associating strings uniquely with scanner/parser information such as keyword type, index in script or function literal pool, etc. Atoms play three roles in JSRef: as literals referred to by unaligned 16-bit immediate bytecode operands, as unique string descriptors for efficient property name hashing, and as members of the root GC set for perfect GC. This design therefore requires atoms to be manually reference counted, from script literal pools (JSAtomMap) and object symbol (JSSymbol) entry keys.

Native objects and methods for arrays, booleans, dates, functions, numbers, and strings are implemented using the JS API and certain internal interfaces used as "fast paths".

In general, errors are signaled by false or unoverloaded-null return values, and are reported using JS_ReportError() or one of its variants by the lowest level in order to provide the most detail. Client code can substitute its own error reporting function and suppress errors, or reflect them into Java or some other runtime system as exceptions, GUI dialogs, etc..

File walk-through (BADLY OUT OF DATE!)

jsapi.c, jsapi.h

The public API to be used by almost all client code.  If your client code can't make do with jsapi.h, and must reach into a friend or private js* file, please let us know so we can extend jsapi.h to include what you need in a fashion that we can support over the long run.

jspubtd.h, jsprvtd.h

These files exist to group struct and scalar typedefs so they can be used everywhere without dragging in struct definitions from N different files. The jspubtd.h file contains public typedefs, and is included by jsapi.h. The jsprvtd.h file contains private typedefs and is included by various .h files that need type names, but not type sizes or declarations.

jsdbgapi.c, jsdbgapi.h

The Debugging API, still very much under development. Provided so far:

jsconfig.h

Various configuration macros defined as 0 or 1 depending on how JS_VERSION is defined (as 10 for JavaScript 1.0, 11 for JavaScript 1.1, etc.). Not all macros are tested around related code yet. In particular, JS 1.0 support is missing from JSRef. JS 1.2 support will appear in a future JSRef release.
 

js.c

The "JS shell", a simple interpreter program that uses the JS API and more than a few internal interfaces (some of these internal interfaces could be replaced by jsapi.h calls). The js program built from this source provides a test vehicle for evaluating scripts and calling functions, trying out new debugger primitives, etc.

jsarray.*, jsbool.*, jdsdate.*, jsfun.*, jsmath.*, jsnum.*, jsstr.*

These file pairs implement the standard classes and (where they exist) their underlying primitive types. They have similar structure, generally starting with class definitions and continuing with internal constructors, finalizers, and helper functions.

jsobj.*, jsscope.*

These two pairs declare and implement the JS object system. All of the following happen here: The details of an object map (scope) are mostly hidden in jsscope.[ch], where scopes start out as linked lists of symbols, and grow after some threshold into PR hash tables.

jsatom.c, jsatom.h

The atom manager. Contains well-known string constants, their atoms, the global atom hash table and related state, the js_Atomize() function that turns a counted string of bytes into an atom, and literal pool (JSAtomMap) methods.

jsgc.c, jsgc.h

[TBD]

jsinterp.*, jscntxt.*

The bytecode interpreter, and related functions such as Call and AllocStack, live in jsinterp.c. The JSContext constructor and destructor are factored out into jscntxt.c for minimal linking when the compiler part of JS is split from the interpreter part into a separate program.

jsemit.*, jsopcode.tbl, jsopcode.*, jsparse.*, jsscan.*, jsscript.*

Compiler and decompiler modules. The jsopcode.tbl file is a C preprocessor source that defines almost everything there is to know about JS bytecodes. See its major comment for how to use it. For now, a debugger will use it and its dependents such as jsopcode.h directly, but over time we intend to extend jsdbgapi.h to hide uninteresting details and provide conveniences. The code generator is split across paragraphs of code in jsparse.c, and the utility methods called on JSCodeGenerator appear in jsemit.c. Source notes generated by jsparse.c and jsemit.c are used in jsscript.c to map line number to program counter and back.

jstypes.h, jslog2.c

Fundamental representation types and utility macros. This file alone among all .h files in JSRef must be included first by .c files. It is not nested in .h files, as other prerequisite .h files generally are, since it is also a direct dependency of most .c files and would be over-included if nested in addition to being directly included. The one "not-quite-a-macro macro" is the JS_CeilingLog2() function in jslog2.c.

jsarena.c, jsarena.h

Last-In-First-Out allocation macros that amortize malloc costs and allow for en-masse freeing. See the paper mentioned in prarena.h's major comment.

jsutil.c, jsutil.h

The JS_ASSERT macro is used throughout JSRef source as a proof device to make invariants and preconditions clear to the reader, and to hold the line during maintenance and evolution against regressions or violations of assumptions that it would be too expensive to test unconditionally at run-time. Certain assertions are followed by run-time tests that cope with assertion failure, but only where I'm too smart or paranoid to believe the assertion will never fail...

jsclist.h

Doubly-linked circular list struct and macros.

jscpucfg.c

This standalone program generates jscpucfg.h, a header file containing bytes per word and other constants that depend on CPU architecture and C compiler type model. It tries to discover most of these constants by running its own experiments on the build host, so if you are cross-compiling, beware.

prdtoa.c, prdtoa.h

David Gay's portable double-precision floating point to string conversion code, with Permission To Use notice included.

prhash.c, prhash.h

Portable, extensible hash tables. These use multiplicative hash for strength reduction over division hash, yet with very good key distribution over power of two table sizes. Collisions resolve via chaining, so each entry burns a malloc and can fragment the heap.

prlong.c, prlong.h

64-bit integer emulation, and compatible macros that use C's long long type where it exists (my last company mapped long long to a 128-bit type, but no real architecture does 128-bit ints yet).

jsosdep.h

Annoying OS dependencies rationalized into a few "feature-test" macros such as JS_HAVE_LONG_LONG.

jsprf.*

Portable, buffer-overrun-resistant sprintf and friends. For no good reason save lack of time, the %e, %f, and %g formats cause your system's native sprintf, rather than JS_dtoa(), to be used. This bug doesn't affect JSRef, because it uses its own JS_dtoa() call in jsnum.c to convert from double to string, but it's a bug that we'll fix later, and one you should be aware of if you intend to use a JS_*printf()  function with your own floating type arguments - various vendor sprintf's mishandle NaN, +/-Inf, and some even print normal floating values inaccurately.

prmjtime.c, prmjtime.h

Time functions. These interfaces are named in a way that makes local vs. universal time confusion likely. Caveat emptor, and we're working on it. To make matters worse, Java (and therefore JavaScript) uses "local" time numbers (offsets from the epoch) in its Date class.

Additional Resources (links, API docs, and newsgroups)