FAQ for Compiler Final Project 2005

1. What are the rules for constant def?
A:   legal_id_name = constant_string or a previously defined constant
     constant strings are either an integer or a float
     float follows the syntax of C except that we need a decimal point

     constant variable is not a normal variable.
     It has r-value, but no l-value. That is, a constant variable be
     on the left hand side of the assignement

2. How to return a function value?
A:   In each function, the name of the function is a variable that
     has l-value, but no r-value. The value of the variable at the end
     of the function execution returns back to the called procedure/function.
     Read page 3 of slide0.pdf for an example.

3. for loop裡的statement我猜是只能放1個statement,
   因為沒有END FOR之類的結束符
   號, 但if裡的也是statement, 所以老師的意思是if, for, while裡面只能有一行嗎??
A: No!
   A statement can be a sequence of statements enclosed by begin end.
   example:
	FOR i := 1 TO 10 DO
	BEGIN
		IF i / 2 * 2 == i
		THEN
			BEGIN
			 x := x + i;
			 x := x * x;
			END
		ENDIF;
		y := x + x;
	END

4. FOR var := int-expression-1 TO int-expression-2 DO
   其中的var要是之前有宣告過的嗎? 還是算是宣告, 可以放新變數?
A: var must be declared before.
   var must be integer.

   Same for FOR ... DOWNTO ... DO

5. 同4, int-expression-1只在進入FOR之前做一次, 但int-expression-2有兩種可能, 
一種是先算好再進入FOR, 一種是每次算, 不知道是哪一種?
A: int-expression-2每次算.

6. In arithmetic expression :    - no auto-type conversion
what does this mean?
A: It means the following.
       1.  int_expr <operator> float_expr    -->  error
       2.  int_var := float_expr    --> error 
       3.  float_var := int_expr    --> error 
   only the followings are correct.
       1. int_expr <operator> int_expr
       2. float_expr <operator> float_expr
       3. int_var := int_expr
       4. float_var := flaot_expr

7. "Can have "(" and ")" in expression"
   what is this ?
A: It means ( and ) and allowed in expressions.
That is,
     expression -> (expression)

8. 投影片中有要求
   check for divide by 0 in compile and run time
   need to check array out of bound in compile and run time
   如果只是檢查如 y := x / 0 這種還好
   萬一 y := x / z 這種也要檢查  是不是就要像interpreter算出 z 的值
   可是如果之前某行在算 z 的時候有syntax error
   這樣 z 算出來的值就可能沒有必定的準則了
   到時候判定 y := x / z 也可能會連帶出現問題
   check array out of bound 也有類似的問題  (如 a[i] 不知道 i 為何值)
A: You will know the value of variables in run time.
So you can generate a code like
        load the value of z into register4
        if value of register4 is zero then goto error_divid_by_zero
Write a C-- code for handling run time error
        error_divide_by_zero:
              error handling code in run time

9. What are the priority of operators?
A:
	The same as C.

10. How to use "not"?
A:
	not logical_expressions

11. 教授有說過浮點數必須要有小數點，那是否小數點前後都必須有數字呢?
因為如果不是的話 ARRAY[1..10] 裡的1應該會parse 成1. (float)
                                 10 應該會parse 成.10 (float)
A: 
小數點前後不須有數字, but at least one side must have digits.
In array declaration, there is no space between the two dots.
That is, ARRAY[1. .10] is illegal.

12.    function的回傳值是以 function name表示
那有沒有可能其 code中出現多次的 function name?
ex: FUNCTION  Foo( x : INTEGER ) : INTEGER
   BEGIN
         Foo := x + 3;
         Foo := x + 5;
         Other Statement;
   END
如果可能且是合法的話，是以最後出現的為準嗎?
又出現再其後的statement是否需要繼續做? 
還是說一看到 Foo 就馬上retuen?
A: A function name is a variable.
So your above example is legal.
You don't immediately perform a return when you see the function name
because it is simply a variable.

13. Type define中是不是也跟constant一樣也是剛宣告過的就可以用?
  Ex: TYPE
      Int1  = INTEGER;
      Int_Array  = ARRAY[1..10]  of Int1;
      ENDTYPE
A: Yes.