[Feature][Materialized-View] support materialized view on vectorized engine #10792
Conversation
github-actions
bot
added
area/planner
BiteTheDDDDt
force-pushed
the
dev_mv
branch
4 times, most recently
from
33600e3 to
9625fc2
Compare
BiteTheDDDDt
force-pushed
the
dev_mv
branch
2 times, most recently
from
0cf4fdd to
48b1149
Compare
github-actions
bot
added
the
area/sql/function
be/src/olap/schema_change.cpp
Outdated
| } else if (_schema_mapping[i].materialized_function == "hll_hash") { | ||
| _do_materialized_transform = hll_hash; | ||
| } else if (_schema_mapping[i].materialized_function == "count_field") { | ||
| } else if (_schema_mapping[i].materialized_function == "is_not_null_pred") { |
There was a problem hiding this comment.
if we update count_field name will cause update BE the function can not process the logic. keep it and remove after next version
be/src/runtime/descriptors.cpp
Outdated
| for (int i = 0; i < row_tuples.size(); ++i) { | ||
| TupleDescriptor* tupleDesc = desc_tbl.get_tuple_descriptor(row_tuples[i]); | ||
|
|
||
| if (skip_delete) { |
There was a problem hiding this comment.
do not treat each slot as a tuple
be/src/vec/functions/is_not_null.cpp
Outdated
|
|
||
| for (size_t i = 0; i < input_rows_count; ++i) { | ||
| res_data[i] = !src_data[i]; | ||
| res_data[i] = src_data[i] == 0; |
| mvColumnName = baseColumnName; | ||
| } else { | ||
| mvColumnName = mvColumnBuilder(functionName, baseColumnName); | ||
| if (!functionChild0.getType().isStringType()) { |
There was a problem hiding this comment.
why the origin do not have the logic
BiteTheDDDDt
force-pushed
the
dev_mv
branch
3 times, most recently
from
65eef10 to
96d242f
Compare
be/src/runtime/descriptors.cpp
Outdated
| init_tuple_idx_map(); | ||
| init_has_varlen_slots(); | ||
|
|
||
| DCHECK(nullable_tuples.size() == _tuple_desc_map.size()) |
There was a problem hiding this comment.
seems no need move the DCHECK to here
| public int getAsInt() { | ||
| pendingMaxColUniqueId++; | ||
| return pendingMaxColUniqueId; | ||
| } |
There was a problem hiding this comment.
1、now in rollup, we don't support light schema change, this is unnecessary.
2、in the end, we also need update tbl max unique id . for example in SchemaChangeJobV2.onFinished() , we do like this:
//update max column unique id
int maxColUniqueId = tbl.getMaxColUniqueId();
for (Column column : tbl.getFullSchema()) {
if (column.getUniqueId() > maxColUniqueId) {
maxColUniqueId = column.getUniqueId();
}
}
tbl.setMaxColUniqueId(maxColUniqueId);
3、i will refact the light schema change feature, here you can just delete this code about generating col_unique_id
update update update update update update update update update update update update fix fe ut update update update update update update update
|
LGTM |
github-actions
bot
added
the
approved
|
PR approved by at least one committer and no changes requested. |
|
PR approved by anyone and no changes requested. |
3 participants
Proposed changes
Issue Number: close #xxx
Problem Summary:
Describe the overview of changes.
Checklist(Required)
Further comments
If this is a relatively large or complex change, kick off the discussion at dev@doris.apache.org by explaining why you chose the solution you did and what alternatives you considered, etc...