Github user JamesRTaylor commented on a diff in the pull request:
https://github.com/apache/phoenix/pull/34#discussion_r23673555
--- Diff:
phoenix-core/src/main/java/org/apache/phoenix/schema/MetaDataClient.java ---
@@ -945,20 +960,40 @@ public MutationState createIndex(CreateIndexStatement
statement, byte[][] splits
PDataType dataType =
MetaDataUtil.getViewIndexIdDataType();
ColumnName colName =
ColumnName.caseSensitiveColumnName(MetaDataUtil.getViewIndexIdColumnName());
allPkColumns.add(new Pair<ColumnName,
SortOrder>(colName, SortOrder.getDefault()));
- columnDefs.add(FACTORY.columnDef(colName,
dataType.getSqlTypeName(), false, null, null, false, SortOrder.getDefault()));
+ columnDefs.add(FACTORY.columnDef(colName,
dataType.getSqlTypeName(), false, null, null, false, SortOrder.getDefault(),
null));
}
// First columns are the indexed ones
- for (Pair<ColumnName, SortOrder> pair : indexedPkColumns) {
- ColumnName colName = pair.getFirst();
- PColumn col = resolver.resolveColumn(null,
colName.getFamilyName(), colName.getColumnName()).getColumn();
- unusedPkColumns.remove(col);
- // Ignore view constants for updatable views as we
don't need these in the index
- if (col.getViewConstant() == null) {
- PDataType dataType =
IndexUtil.getIndexColumnDataType(col);
- colName =
ColumnName.caseSensitiveColumnName(IndexUtil.getIndexColumnName(col));
- allPkColumns.add(new Pair<ColumnName,
SortOrder>(colName, pair.getSecond()));
- columnDefs.add(FACTORY.columnDef(colName,
dataType.getSqlTypeName(), col.isNullable(), col.getMaxLength(),
col.getScale(), false, SortOrder.getDefault()));
- }
+ for (Pair<ParseNode, SortOrder> pair :
indexParseNodeAndSortOrderList) {
+ ParseNode parseNode = pair.getFirst();
+ ColumnName colName = null;
+ // if it is a column
+ if (parseNode instanceof ColumnParseNode) {
--- End diff --
Or a better way is to change unusedPkColumns to be a Set<Expression>. These
can be created directly through
List<PColumn> pkColumns = table.getPKColumns();
List<Expression> unusedPkColumns = new LinkedHashSet(pkColumns.size());
for (int i = 0; i < pkColumns; i++) {
unusedPkColumns.add(new RowKeyColumnExpression(pkColumns.get(i),
new RowKeyValueAccessor(pkColumns, i));
}
Then, here you can get rid of the special case, and instead after
compilation, do the following:
if (expression.isStateless()) { // Will be true for any constant
(including a view constant)
continue;
}
unusedPkColumns.remove(expression);
We don't need to include an expression that evaluates to a constant in the
index. You can test this by adding a function index on something like 1 + 2.
In this way, you can remove the if statement completely.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
