1 Review

2 Creating a Runnable Program  What is the function of the compiler?  What is the function of the linker?  Java doesn't have a linker. If the "main" class uses other classes, what happens? does javac combine all classes in one class file? does java load the classes? when?  what does this statement do? Class.forName("com.mysql.jdbc.Driver");

3 Linker  The linker "resolves" external references, it fills in the missing values in object files.  But, it doesn't check that the names match the expected type! Example: homework 6A problem 1: /* File 1 */ char *foo; extern double bar;... printf("%s", foo); printf("%f", bar); /* File 2 */ extern double foo; char* bar = "foo!";... printf("%d", foo); printf("%s", bar);

4 Homework 4 There are two parts to a language:  syntax is the grammar rules for writing (or saying) the language. It describes what are legal expressions. The grammar has 2 parts. lexical grammar - rules for tokens (words, punctuation, space) syntactic grammar - rules for statements  What notation is commonly used for these two parts?  semantics describes the meaning of the language usually given in a manual (in human language) formal semantics: axiomatic, denotational, and operational semantics precisely describe effect of language on the computer's "state", can be used to prove program correctness.

5 Homework 5 Representation of primitive data types:  how big is an int, long, float, double?  what's the largest value you can store in a long?  what format is used to store signed integers?  If you look at a (signed) integer in memory and see: what value does this represent?  If you look in memory using hexadecimal format and see: 7FFFFFFF is this a positive or negative value? What's the value?

6 Homework 5 Representation of primitive data types:  floating point numbers are stored as 3 components. What are the 3 components?  the largest float is about 2 x ( bias ), so the smallest value should be 1 x ( 0 - bias ), but its not!  why is the smallest float 1 x ( 1.4E-45 ). Why?  what floating point is returned in these cases, using IEEE standard floating point: float x = 1.0E20; float result1 = x*x*x*x; float result2 = 1.0/(x*x*x*x); double zero = 0.0; double result3 = 1.0/zero; double result4 = zero * result1;  In Java, how can you test if x (say, "double x") is Infinity?  Test for x is NaN?

7 Test for Infinity and NaN In Java... double x =...; if ( x == Double.POSITIVE_INFINITY )... if ( Double.isInfinite(x) && x > 0 )... if ( x == Double.NEGATIVE_INFINITY )... if ( Double.isInfinite(x) && x < 0 )... /* for NaN there is no constant. Use function */ if ( Double.isNaN(x) )... In C++ use: #include numeric_limits dbl; if ( x == dbl.infinity( ) )...

8 3 Memory areas  The memory area allocated for a program's data is divided into 3 areas. What are their names?

9 3 Memory areas  The memory area allocated for a program's data is divided into 3 areas. What are their names? Static area Stack Heap  How are these three areas used?  What sort of values are placed there? depends on the language: statically typed languages generally use them as discussed in class dynamic typed languages (Perl, SmallTalk, Scheme) use the heap for most things

10 Weird functions  A programmer writes a timer like this: time_t tstart = time(0); // starting time doSomeWork( ); time_t tstop = time(0); // stopping time /* now print when the program started/stopped */ char* start = ctime( tstart ); char* stop = ctime( tstop ); printf("Job started: %s\n", start); printf("Job finished: %s\n", stop); Job started: Wed Dec :14:39 PM Job finished: Wed Dec :14:39 PM  but both output values are always the same! Why?

11 Weird functions: static memory  ctime uses a static buffer to create a string of the formatted time ("Wed Dec 14...").  it returns a pointer to this buffer.  it reuses the same buffer each time it is invoked! char* start = ctime(time1); Tue Dec :00 AM char* start Wed Dec :00 PM char* stop = ctime(time2); Static buffer:

12 Weird functions: why static memory?  Why use static memory for the string buffer?  why not a local (stack automatic) variable? ctime( time_t time ) { char buf[80];... return buf; Wed Dec :00:00 PM\n char* time = ctime(time1);... x = fun(10,20,30);... printf("%s\n", time); stack:

13 Weird functions: why static memory?  why not use malloc (heap dynamic) variable? ctime( time_t time ) { char* buf = (char *)malloc(80);... return buf; Wed Dec :00 AM Wed Dec :00 AM Wed Dec :00 PM char* time; time = ctime(time1);... time = ctime(time2);... time = ctime(time3); heap:

14 Alignment  as described in class, int, float, and double values must usually be aligned on word boundaries (word = 4 bytes)  some CPU require "double" aligned on double word boundaries (8 bytes).  how much space is used by these struct variables? struct one { char c; double d; } x; /* try changing order */ struct two { double d; char c; } y; struct three { char c1; double d1; char c2; double d2; char c3; double d3 } z;

15 Data type compatibility  each language has rules for compatibility of data types  C++, Java, C#: "widening" conversions allowed; others require a cast and may throw an exception.  what about compatibility of user defined types? double dbl = 3.5; float flt = 1.2F; int nint = ; long nlng = ; /* OK */ dbl = flt; dbl = nlng; dbl = nint; flt = nlng; flt = nint; nlng = int; /* Error (but OK in C) */ flt = dbl; nint = nlng; nlng = float; is there a problem here?

16 Type Compatibility for user types struct A { float x; char c; }; struct B { float x; char c; }; struct C { float z; char c; }; struct D { char c; float x; }; int main( ) { struct A a; struct B b; struct C c; struct D d; a.x = 0.5; a.c = 'a'; b = a; // ?? c = a; // ?? d = a; // ?? if (b == a) // true? Which of these assignments are allowed? Is the test condition true?

17 Constants  Manifest constant (literal): const int BUFSIZE = 1024; char buf[BUFSIZE];  Compile time (everything is known): const int BUFSIZE = 8 * 128;  Load time (determined when program is loaded): static final String user = System.getenv("username"); /* java */  Run-time: int fun(const int n) { const size = n*n;... } final int width = 400; /* java instance var */

18 Scope of Names  Categories of scope rules: lexical (static) scope - scope known by looking at the code. dynamic scope - follows flow of execution  most languages use lexical scope, but each has its own rules about redefining names in inner scopes.  example of name conflict: #include /* hypot is an error in C but not in C++ ! */ float hypot(float x, float y) { return sqrt(x*x + y*y); }

19 Keeping Track of Names  Compiler (or interpreter) uses a symbol table to record known names and their attributes (bindings).  For lexical scope, there are two approaches: a new symbol table for each scope: a stack of symbol tables a linked list of definitions for each symbol (name), elements of the linked list have scope number to keep track of active scopes: scope stack  Symbol table is not needed after program is linked, but is usually hidden in the executable code to aid debugging.  To make program smaller or inhibit reverse engineering: strip  C#: "Dotfuscation" pun on "obfuscation" - to make confused or opaque