LogRecordIterator uses a LogIterator, which does the actual iteration.

HelloWorld wrote:

cysun wrote:

LogRecordIterator uses a LogIterator, which does the actual iteration.

I still don't understand on how it exactly works.. For example:

42 a b 0 c d 14 e f 28

(Figure 13-7, Page 354)

   public BasicLogRecord next() {   
      if (currentrec == 0)
         moveToNextBlock();
      currentrec = pg.getInt(currentrec);
      return new BasicLogRecord(pg, currentrec+INT_SIZE);
   }

what is the currentrec? I thought it's the pointer position along the block, where block is divided into 4 bytes pages, which is the INT_SIZE but i'm not sure.. when is it that it reaches moveToNextBlock(); in this example? where does it move next? when does it read the "a b", "c d", etc..??

Thanks

I think Figure 13-7 is wrong - it shows that the position of the last record is after the current record, while I think it should be before the current record. If you use the example in the lecture notes instead, you should be able to follow the LogIterator code.

HelloWorld wrote:

Hmm.. So I'm using the example of your lecture note:

16 0 1 Hi 4 2 32

i understand how it works, 16 will moves it to position 16, then get the content of the 16, which is 4, and then i'm assuming that it should read 2 and 32, but i still don't get this part:

• why it returns new BasicLogRecord(pg, currentrec+INT_SIZE);? since INT_SIZE is 4, what does it mean?
• what happened on:
        if (currentrec == 0)
           moveToNextBlock();
  after content is 0, then move to the NextBlock(), which in this example is?

Thanks

OK, I think my example is wrong, and the one in the textbook is correct. Here's how the example in the textbook works:

1. When a new block is loaded in memory, read from position 0, so currentrec=42.

2. When next() is called, read an int from currentrec position so currentrec=28, then return the position currentrec+INT_SIZE, i.e. the starting position of ["e","f"].

3. When next() is called again, read an int from currentrec position so currentrec=14, then return the position currentrec+INT_SIZE, i.e. the starting position of ["c","d"].

4. When next() is called again, read an int from currentrec position so currentrec=0, then return the position currentrec+INT_SIZE, i.e. the starting position of ["a","b"].

5. When next() is called again, currentrec is 0 which means all the records in the page have been read, so load another block.

BasicLogRecord is simply a block and the starting position of a record in the block. It's up to the recovery manager to read and interpret the record - see the next() method of LogRecordIterator.