gortiz commented on code in PR #11453:
URL: https://github.com/apache/pinot/pull/11453#discussion_r1308291268
##########
pinot-core/src/main/java/org/apache/pinot/core/common/datablock/DataBlockBuilder.java:
##########
@@ -107,144 +104,83 @@ public void setNullRowIds(@Nullable RoaringBitmap
nullRowIds)
public static RowDataBlock buildFromRows(List<Object[]> rows, DataSchema
dataSchema)
throws IOException {
- DataBlockBuilder rowBuilder = new DataBlockBuilder(dataSchema,
DataBlock.Type.ROW, rows.size());
+ int numRows = rows.size();
+ DataBlockBuilder rowBuilder = new DataBlockBuilder(dataSchema,
DataBlock.Type.ROW, numRows);
// TODO: consolidate these null utils into data table utils.
// Selection / Agg / Distinct all have similar code.
- int numColumns = rowBuilder._numColumns;
+ ColumnDataType[] storedTypes = dataSchema.getStoredColumnDataTypes();
+ int numColumns = storedTypes.length;
RoaringBitmap[] nullBitmaps = new RoaringBitmap[numColumns];
- DataSchema.ColumnDataType[] columnDataTypes =
dataSchema.getColumnDataTypes();
- DataSchema.ColumnDataType[] storedColumnDataTypes =
dataSchema.getStoredColumnDataTypes();
Object[] nullPlaceholders = new Object[numColumns];
for (int colId = 0; colId < numColumns; colId++) {
nullBitmaps[colId] = new RoaringBitmap();
- nullPlaceholders[colId] =
columnDataTypes[colId].convert(storedColumnDataTypes[colId].getNullPlaceholder());
+ nullPlaceholders[colId] = storedTypes[colId].getNullPlaceholder();
}
- rowBuilder._numRows = rows.size();
ByteBuffer byteBuffer = ByteBuffer.allocate(rowBuilder._rowSizeInBytes);
- for (int rowId = 0; rowId < rows.size(); rowId++) {
+ for (int rowId = 0; rowId < numRows; rowId++) {
byteBuffer.clear();
Object[] row = rows.get(rowId);
- for (int colId = 0; colId < rowBuilder._numColumns; colId++) {
+ for (int colId = 0; colId < numColumns; colId++) {
Object value = row[colId];
if (value == null) {
nullBitmaps[colId].add(rowId);
value = nullPlaceholders[colId];
}
- switch (rowBuilder._columnDataTypes[colId]) {
+ switch (storedTypes[colId]) {
// Single-value column
case INT:
- byteBuffer.putInt(((Number) value).intValue());
+ byteBuffer.putInt((int) value);
Review Comment:
Is there any difference here? Just that the casting is going to be more
restrictive, right? I mean, performance should be the same but now it would
fail if value is not exactly an Integer
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
