mirror of
https://github.com/hanzoai/datastore.git
synced 2026-09-26 13:55:07 +00:00
avogarandClaude Opus 4.6
0243158d48
Non-adjacent grouping could reorder predicates past intentional guards, e.g. `PREWHERE tags['safe'] != '' AND value > 0 AND toUInt64(tags['unsafe']) > 0` would pull the throwing conversion before the `value > 0` filter. Revert to adjacent-only grouping in `tryBuildPrewhereSteps`. This is safe because `MergeTreeWhereOptimizer` already groups subcolumns of the same storage column together, so they arrive adjacent for the WHERE-to-PREWHERE path. Co-Authored-By: Claude Opus 4.6 <[email protected]>
147 lines
6.8 KiB
XML
147 lines
6.8 KiB
XML
<test>
|
|
<!-- Benchmarks Map subcolumn reads with optimize_functions_to_subcolumns enabled.
|
|
When multiple Map subcolumn conditions appear in PREWHERE, they should be grouped
|
|
into a single read step to avoid redundant deserialization of the same Map data.
|
|
Tests SELECT of multiple keys, count() with WHERE, and count() with PREWHERE
|
|
for both Wide and Compact part formats. -->
|
|
|
|
<settings>
|
|
<max_insert_threads>4</max_insert_threads>
|
|
<use_query_condition_cache>0</use_query_condition_cache>
|
|
</settings>
|
|
|
|
<!-- Wide part format -->
|
|
<create_query>
|
|
CREATE TABLE map_subcol_pw_wide (id UInt64, tags Map(String, String))
|
|
ENGINE = MergeTree ORDER BY id
|
|
SETTINGS min_bytes_for_wide_part = 0, min_rows_for_wide_part = 0
|
|
</create_query>
|
|
|
|
<!-- Compact part format -->
|
|
<create_query>
|
|
CREATE TABLE map_subcol_pw_compact (id UInt64, tags Map(String, String))
|
|
ENGINE = MergeTree ORDER BY id
|
|
SETTINGS min_bytes_for_wide_part = '10G', min_rows_for_wide_part = 1000000000
|
|
</create_query>
|
|
|
|
<fill_query>
|
|
INSERT INTO map_subcol_pw_wide
|
|
SELECT number AS id,
|
|
mapFromArrays(
|
|
arrayMap(i -> 'k' || toString(i), range(16)),
|
|
arrayMap(i -> toString(number + i), range(16))) AS tags
|
|
FROM numbers_mt(5000000)
|
|
</fill_query>
|
|
<fill_query>OPTIMIZE TABLE map_subcol_pw_wide FINAL SETTINGS mutations_sync=1</fill_query>
|
|
|
|
<fill_query>
|
|
INSERT INTO map_subcol_pw_compact
|
|
SELECT number AS id,
|
|
mapFromArrays(
|
|
arrayMap(i -> 'k' || toString(i), range(16)),
|
|
arrayMap(i -> toString(number + i), range(16))) AS tags
|
|
FROM numbers_mt(5000000)
|
|
</fill_query>
|
|
<fill_query>OPTIMIZE TABLE map_subcol_pw_compact FINAL SETTINGS mutations_sync=1</fill_query>
|
|
|
|
<!-- Wide: SELECT multiple keys -->
|
|
<query>
|
|
SELECT tags['k0'], tags['k1'], tags['k2'], tags['k3'],
|
|
tags['k4'], tags['k5'], tags['k6'], tags['k7'],
|
|
tags['k8'], tags['k9'], tags['k10'], tags['k11'],
|
|
tags['k12'], tags['k13'], tags['k14'], tags['k15']
|
|
FROM map_subcol_pw_wide
|
|
FORMAT Null
|
|
SETTINGS optimize_functions_to_subcolumns=1
|
|
</query>
|
|
|
|
<!-- Compact: SELECT multiple keys -->
|
|
<query>
|
|
SELECT tags['k0'], tags['k1'], tags['k2'], tags['k3'],
|
|
tags['k4'], tags['k5'], tags['k6'], tags['k7'],
|
|
tags['k8'], tags['k9'], tags['k10'], tags['k11'],
|
|
tags['k12'], tags['k13'], tags['k14'], tags['k15']
|
|
FROM map_subcol_pw_compact
|
|
FORMAT Null
|
|
SETTINGS optimize_functions_to_subcolumns=1
|
|
</query>
|
|
|
|
<!-- Wide: count() with WHERE -->
|
|
<query>
|
|
SELECT count() FROM map_subcol_pw_wide
|
|
WHERE tags['k0'] != '' AND tags['k1'] != '' AND tags['k2'] != '' AND tags['k3'] != ''
|
|
AND tags['k4'] != '' AND tags['k5'] != '' AND tags['k6'] != '' AND tags['k7'] != ''
|
|
AND tags['k8'] != '' AND tags['k9'] != '' AND tags['k10'] != '' AND tags['k11'] != ''
|
|
AND tags['k12'] != '' AND tags['k13'] != '' AND tags['k14'] != '' AND tags['k15'] != ''
|
|
SETTINGS optimize_functions_to_subcolumns=1, optimize_move_to_prewhere=0
|
|
</query>
|
|
|
|
<!-- Compact: count() with WHERE -->
|
|
<query>
|
|
SELECT count() FROM map_subcol_pw_compact
|
|
WHERE tags['k0'] != '' AND tags['k1'] != '' AND tags['k2'] != '' AND tags['k3'] != ''
|
|
AND tags['k4'] != '' AND tags['k5'] != '' AND tags['k6'] != '' AND tags['k7'] != ''
|
|
AND tags['k8'] != '' AND tags['k9'] != '' AND tags['k10'] != '' AND tags['k11'] != ''
|
|
AND tags['k12'] != '' AND tags['k13'] != '' AND tags['k14'] != '' AND tags['k15'] != ''
|
|
SETTINGS optimize_functions_to_subcolumns=1, optimize_move_to_prewhere=0
|
|
</query>
|
|
|
|
<!-- Wide: count() with PREWHERE -->
|
|
<query>
|
|
SELECT count() FROM map_subcol_pw_wide
|
|
PREWHERE tags['k0'] != '' AND tags['k1'] != '' AND tags['k2'] != '' AND tags['k3'] != ''
|
|
AND tags['k4'] != '' AND tags['k5'] != '' AND tags['k6'] != '' AND tags['k7'] != ''
|
|
AND tags['k8'] != '' AND tags['k9'] != '' AND tags['k10'] != '' AND tags['k11'] != ''
|
|
AND tags['k12'] != '' AND tags['k13'] != '' AND tags['k14'] != '' AND tags['k15'] != ''
|
|
SETTINGS optimize_functions_to_subcolumns=1
|
|
</query>
|
|
|
|
<!-- Compact: count() with PREWHERE -->
|
|
<query>
|
|
SELECT count() FROM map_subcol_pw_compact
|
|
PREWHERE tags['k0'] != '' AND tags['k1'] != '' AND tags['k2'] != '' AND tags['k3'] != ''
|
|
AND tags['k4'] != '' AND tags['k5'] != '' AND tags['k6'] != '' AND tags['k7'] != ''
|
|
AND tags['k8'] != '' AND tags['k9'] != '' AND tags['k10'] != '' AND tags['k11'] != ''
|
|
AND tags['k12'] != '' AND tags['k13'] != '' AND tags['k14'] != '' AND tags['k15'] != ''
|
|
SETTINGS optimize_functions_to_subcolumns=1
|
|
</query>
|
|
|
|
<!-- Table with an extra column to test WHERE optimizer grouping.
|
|
When Map subcolumn conditions are interleaved with another column in WHERE,
|
|
the optimizer should group Map conditions together to avoid redundant reads. -->
|
|
<create_query>
|
|
CREATE TABLE map_subcol_pw_mixed (id UInt64, tags Map(String, String), value UInt32)
|
|
ENGINE = MergeTree ORDER BY id
|
|
SETTINGS min_bytes_for_wide_part = 0, min_rows_for_wide_part = 0
|
|
</create_query>
|
|
|
|
<fill_query>
|
|
INSERT INTO map_subcol_pw_mixed
|
|
SELECT number AS id,
|
|
mapFromArrays(
|
|
arrayMap(i -> 'k' || toString(i), range(16)),
|
|
arrayMap(i -> toString(number + i), range(16))) AS tags,
|
|
number AS value
|
|
FROM numbers_mt(5000000)
|
|
</fill_query>
|
|
<fill_query>OPTIMIZE TABLE map_subcol_pw_mixed FINAL SETTINGS mutations_sync=1</fill_query>
|
|
|
|
<!-- WHERE with interleaved Map subcolumns and another column.
|
|
The optimizer should group all Map conditions together even though
|
|
the 'value' condition appears between them in the original WHERE. -->
|
|
<query>
|
|
SELECT count() FROM map_subcol_pw_mixed
|
|
WHERE tags['k0'] != '' AND tags['k1'] != '' AND tags['k2'] != '' AND tags['k3'] != ''
|
|
AND value > 0
|
|
AND tags['k4'] != '' AND tags['k5'] != '' AND tags['k6'] != '' AND tags['k7'] != ''
|
|
AND tags['k8'] != '' AND tags['k9'] != '' AND tags['k10'] != '' AND tags['k11'] != ''
|
|
AND tags['k12'] != '' AND tags['k13'] != '' AND tags['k14'] != '' AND tags['k15'] != ''
|
|
SETTINGS optimize_functions_to_subcolumns=1, move_all_conditions_to_prewhere=1
|
|
</query>
|
|
|
|
|
|
<drop_query>DROP TABLE IF EXISTS map_subcol_pw_wide</drop_query>
|
|
<drop_query>DROP TABLE IF EXISTS map_subcol_pw_compact</drop_query>
|
|
<drop_query>DROP TABLE IF EXISTS map_subcol_pw_mixed</drop_query>
|
|
</test>
|