2.3.4.4 Row Matrix
The Row Matrix contains the actual data for the rows and columns of the TC. The data is physically arranged in rows; each row contains the data for each of its columns. Each row of column data in the Row Matrix is of the same size and is arranged in the same layout, and the size of each row is specified in the rgib[TCI_bm] value in the TCINFO header structure.
However, in many cases, the Row Matrix is larger than 8 kilobytes and therefore cannot fit in a single data block, which means that a data tree is used to store the Row Matrix in separate data blocks. This means that the row data is partitioned across two or more data blocks and needs special handling considerations.
The design of a TC dictates that each data block MUST store an integral number of rows, which means that rows cannot span across two blocks, and that each block MUST start with a fresh row. This also means that in order for a client to access a particular row in the Row Matrix, it first calculates how many rows fit in a block, and calculates the row index within that block at which the row data is located. The general formulas to calculate the block index and row index for the Nth row are as follows:
Rows per block = Floor((sizeof(block) – sizeof(BLOCKTRAILER)) / TCINFO.rgib[TCI_bm])
Block index = N / (rows per block)
Row index = N % (rows per block)
Each block except the last one MUST have a size of 8192 bytes. If not, the file is considered corrupted. The size of a block is specified in the formula by sizeof(block).
The following diagram illustrates how the data in the Row Matrix is organized.
Figure 8: Data organization of the Row Matrix
In addition to showing the data organization of the Row Matrix, this diagram also illustrates how the rows in the RowIndex relate to the row data in the Row Matrix. As illustrated by the crossing of dotted lines between the two structures, the Row Matrix data is unsorted, which makes searching inefficient. The RowIndex, which is implemented using an embedded BTH indexed by dwRowID, provides the primary search key to lookup specific rows in the Row Matrix.
It is also worth noting that because of the fact that partial rows are not allowed, there might be unused space at the end of the data block (shaded in gray in the diagram). Readers MUST ignore any such "dead space" and MUST NOT interpret its contents.