這章是programming的核心所在。

Conditional Statements

Review Section
This section has described the following points about conditional compilation statements:
1. The PL/SQL language supports single-branching and multiple-branching statements without fall-through, and multiple-branching statements use either the ELSIF or CASE statement.
2. You can use conjunction (AND logic) or inclusion (OR logic) in conditional statements.
3. The PL/SQL language implements three-valued logic with the possibility of true, false, and null.
4. The NVL statement lets you reduce three-valued logic problems to two-valued logic in conditional statements.
5. You need to assign default values to dynamic variables when comparison values may fail because they hold null values.
6. The PL/SQL language supports both simple and searched CASE statements.
7. Searched CASE statements use a default selector of true but can be configured to use false as the selector value.
8. Conditional compilation supports any number of conditional compilation flags, like the $$DEBUG example used in this chapter.
Conditional compilation also supports predefined inquiry directives, like $$PLSQL_LINE and $$PLSQL_UNIT.

Iterative Statements

Review Section
This section has described the following points about iterative statements:
1. Iterative statements are blocks that let you repeat a statement or set of statements.
2. Iterative statements implement an exit guard and a block of repeatable statements.
3. The exit guard compares a loop index or variable against a sentinel value to determine when to exit the loop.
4. Entry guards and exit guards can work to bar entry to or exit from a loop, and they can be intrinsically linked to index counter and sentinel values.
5. To avoid runtime errors, programs should safeguard any dynamic variable values. You do that by checking whether they have null values before your program uses them. This type of checking is necessary because null values don't work properly as index counters because you can't increment them, or as comparison variables because you can't compare them.
6. The CONTINUE, CONTINUE WHEN, and GOTO statements let you skip the balance of iteration through a loop, and they require specialized handling with guard-on-entry loops to avoid infinite loops.
7. The range FOR loop lets you increment or decrement across data by comparing a range of values, and it hides the complexity of its guard-on-entry loop internals.
8. The WHILE loop is a guard-on-entry loop and requires careful attention to avoid infinite loops when skipping iterations inside of its logic.

Cursor Structures

Review Section
This section has described the following points about cursor structures:
1. Cursor structures return row-by-row managed result sets from SQL SELECT statements.
2. Implicit cursors exist for all DML statements, such as the INSERT, UPDATE, DELETE, and MERGE statements.
3. PL/SQL supports the %FOUND, %NOTFOUND, %ISOPEN, and %ROWCOUNT implicit cursor attributes.
4. The SELECT-INTO statement is a single-row implicit cursor.
5. The INSERT, UPDATE, DELETE, and MERGE statements are multiple-row implicit cursors, although you can limit them to a single row.
6. Cursors can be static or dynamic, and dynamic implicit cursors can include placeholders that are references to local variables.
7. SELECT statement cursors can have formal parameters, like functions and procedures.
8. Cursors support nested cursors, which are called subcursors

Bulk Statments

Review Section
This section has described the following points about bulk statements:
1. Bulk processing can work with parallel collections of scalar data types or collections of record data types.
2. The BULK COLLECT statement lets you gather rows from a cursor into a collection, and collections can be single or parallel collections of scalar data types, or collections of records.
3. PL/SQL supports the %BULK_EXCEPTIONS(i) and %BULK_ROWCOUNTS(i) bulk collection attributes.
4. The FORALL statement lets you take a collection and pass it to an INSERT, UPDATE, or DELETE statement.
5. The %ROWCOUNT attribute also works with bulk inserts, updates, and deletes, but in the case of MERGE statements, you're never sure which rows are inserted or updated.
6. Record collections work best when FORALL statements work with INSERT or UPDATE statements.
7. Single scalar collections work best when FORALL statements work with DELETE statements.
8.  By default, Oracle's exception handling stops program logic with a single error, which isn't optimal for bulk processing. The %BULK_EXCEPTION collection attribute lets you override the default exception handling process in a FORALL statement, and lets you capture errors for problem rows while successfully processing those rows without errors.

Mastery Check

The mastery check is a series of true-or-false and multiple-choice questions that let you confirm
how well you understand the material in the chapter. You may check Appendix I for answers to
these questions.

True or False:
1. Conjunctive logic involves determining when two or more things are true at the same time.
True. Conjunctive logic involves two or more comparisons joined by an AND keyword.
Two or all of the comparisons must be true for a conjunctive statement to be true.
2. Inclusion logic involves determining when one or another thing is true at any time.
True. Inclusion logic involves two or more comparisons joined by an OR keyword. At least one item must be found true, and when one is found true, short-circuit logic stops making any remaining comparisons.
3. Short-circuit logic occurs with inclusion logic.
True. Inclusion logic uses short-circuit logic, which stops making comparisons once one item is found true.
4. Databases always rely on two-valued logic.
False. Databases rely on three-valued logic because they must be capable of comparing a null value. That’s why Oracle Database and other databases support the IS [NOT] NULL reference comparison operator.
5. A searched CASE statement may use a string or numeric selector.
True. A searched CASE statement evaluates Boolean logic, or the result of comparison operations.
6. A simple CASE statement can use a numeric selector.
True. A simple CASE statement evaluates a numeric value, and when it finds a match in one of the WHEN clauses, it exits the CASE statement.
7. Conditional compilation supports conditional compilation flags.
True. Conditional compilation supports any number of compilation flags.
8. A CONTINUE statement lets you skip the balance of an iteration through a loop.
True. The CONTINUE statement instructs the program to skip the balance of an iteration cycle and return to the top of the loop.
9. A SELECT-INTO statement is an example of an explicit cursor.
False. A SELECT-INTO statement is an implicit cursor.
10. The FORALL statement lets you perform bulk INSERT statements.
True. The FORLL statement is the key structure for performing bulk INSERT and UPDATE statements.

Multiple Choice:
11. A conditional statement applied against two operands can evaluate which of the following? (Multiple answers possible)
A. The truth of a comparison involving only not-null values
B. The non-truth (or falsity) of a comparison involving only not-null values
C. The truth of a comparison involving one or more null values
D. The non-truth (or falsity) of a comparison involving one or more null values
E. The truth of null values
A is correct. Comparison operations are three-valued logic comparisons unless you’re checking for a null value. Checking for a null value is an evaluation of what a reference points to, not the value that it holds. Comparison operations between two operands only work when both operands hold not-null values.
12. Which of the following are only guard-on-entry loops? (Multiple answers possible)
A. A simple range loop
B. A range FOR loop
C. A WHILE loop
D. A DO-UNTIL loop
E. A DO-WHILE loop
A, B, and C are correct. A simple loop can be coded to guard on entry or exit. A range FOR loop guards on entry and exit, and a WHILE loop guards entry only. DO-UNTIL and DO-WHILE aren’t loop structures in PL/SQL.
13. Which of the following guards entry and exit to the loop in PL/SQL? (Multiple answers possible)
A. A range FOR loop
B. A cursor FOR loop
C. A simple loop
D. A DO-WHILE loop
E. A WHILE loop
A and B are correct. The FOR loop guards entry and exit and manages them implicitly.
14. Which of the following are only guard-on-exit loops? (Multiple answers possible)
A. A simple cursor loop
B. A simple range loop
C. A cursor FOR loop
D. A WHILE loop
E. A range FOR loop
A and B are correct. Only the simple loop allows unfettered access, and allows you to guard on exit. That lets you run the logic once and exit regardless of any conditions before entering the loop.

15. Which of the following collections work best with a bulk delete operation on a welldefined (or normalized) table with a surrogate key for its single-column primary key?
(Multiple answers possible)
A. Parallel scalar collections
B. A single scalar collection
C. A single record collection
D. All of the above
E. None of the above
B and C are correct. Bulk operations only work when there’s one collection to traverse, and that means a single scalar or record collection.