adelapena commented on code in PR #2110:
URL: https://github.com/apache/cassandra/pull/2110#discussion_r1101451079
##########
src/java/org/apache/cassandra/schema/SchemaKeyspace.java:
##########
@@ -688,15 +695,58 @@ private static void
addColumnToSchemaMutation(TableMetadata table, ColumnMetadat
{
AbstractType<?> type = column.type;
if (type instanceof ReversedType)
- type = ((ReversedType) type).baseType;
+ type = ((ReversedType<?>) type).baseType;
- builder.update(Columns)
+ Row.SimpleBuilder rowBuilder = builder.update(Columns)
.row(table.name, column.name.toString())
.add("column_name_bytes", column.name.bytes)
.add("kind", column.kind.toString().toLowerCase())
.add("position", column.position())
.add("clustering_order",
column.clusteringOrder().toString().toLowerCase())
.add("type", type.asCQL3Type().toString());
+
+ // Dynamic data masking functions shouldn't be attached to columns
during rolling upgrades
+ // to avoid sending mutations with columns that are unknown to the old
nodes.
+ ColumnMask mask = column.getMask();
+ if (ColumnMask.clusterSupportsMaskedColumns())
+ {
+ if (mask == null)
+ {
+ rowBuilder.delete("mask_keyspace")
Review Comment:
In other cases yes, but here I would prefer to keep the style of
`SchemaKeyspace`, where column names are always referred to with direct strings
instead of constants. There are so many of these that probably the class would
lose readability if we made all constants.
--
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]
