Make PageProcessor preserve input block laziness #1075

sopel39 · 2019-07-03T06:02:55Z

Followup:

port semi-join to WorkProcessorOperator (@Praveen2112 ?)
port FilterAndProjectOperator to WorkProcessorOperator
This way we get (node-local) dynamic filtering for semi joins and IN (subquery), NOT IN (subquery), x >= ALL (subquery), etc expressions

ScanFilterAndProject can have different source node ID and plan node ID.

sopel39 · 2019-07-08T10:11:33Z

Benchmarks (no perf regression for non-lazy pages).
Benchmarks were done using plain input reference expressions:

AFTER:

Benchmark                                                      (columnCount)  (dictionaryBlocks)  (positionsPerPage)   (type)  Mode  Cnt    Score   Error  Units
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              2               false                  32  varchar  avgt   50  141.904 ± 0.821  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              2               false                  32   bigint  avgt   50   45.060 ± 0.567  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              2               false                1024  varchar  avgt   50   77.402 ± 0.540  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              2               false                1024   bigint  avgt   50    9.260 ± 0.039  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              2                true                  32  varchar  avgt   50   90.962 ± 0.611  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              2                true                  32   bigint  avgt   50   56.806 ± 0.832  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              2                true                1024  varchar  avgt   50   19.393 ± 0.171  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              2                true                1024   bigint  avgt   50    8.482 ± 0.017  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              4               false                  32  varchar  avgt   50  211.098 ± 0.377  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              4               false                  32   bigint  avgt   50   72.684 ± 0.713  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              4               false                1024  varchar  avgt   50   96.418 ± 0.502  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              4               false                1024   bigint  avgt   50   13.353 ± 0.073  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              4                true                  32  varchar  avgt   50  149.502 ± 0.675  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              4                true                  32   bigint  avgt   50   94.367 ± 1.635  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              4                true                1024  varchar  avgt   50   24.067 ± 0.242  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              4                true                1024   bigint  avgt   50   12.741 ± 0.066  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              8               false                  32   bigint  avgt   50  125.515 ± 1.271  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              8               false                1024  varchar  avgt   50  130.306 ± 0.741  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              8               false                1024   bigint  avgt   50   21.516 ± 0.048  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              8                true                  32  varchar  avgt   50  256.790 ± 1.033  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              8                true                  32   bigint  avgt   50  162.861 ± 2.340  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              8                true                1024  varchar  avgt   50   33.036 ± 0.096  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              8                true                1024   bigint  avgt   50   21.007 ± 0.081  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             16               false                  32   bigint  avgt   50  224.977 ± 1.399  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             16               false                1024  varchar  avgt   50  200.095 ± 0.717  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             16               false                1024   bigint  avgt   50   37.257 ± 0.113  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             16                true                  32  varchar  avgt   50  484.378 ± 1.285  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             16                true                  32   bigint  avgt   50  292.979 ± 1.243  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             16                true                1024  varchar  avgt   50   50.663 ± 0.503  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             16                true                1024   bigint  avgt   50   37.043 ± 0.152  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             32               false                  32   bigint  avgt   50  420.169 ± 2.380  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             32               false                1024  varchar  avgt   50  342.290 ± 1.434  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             32               false                1024   bigint  avgt   50   69.130 ± 0.343  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             32                true                  32   bigint  avgt   50  612.408 ± 4.062  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             32                true                1024  varchar  avgt   50   87.303 ± 0.618  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             32                true                1024   bigint  avgt   50   71.148 ± 0.465  ms/op

BEFORE:

Benchmark                                                      (columnCount)  (dictionaryBlocks)  (positionsPerPage)   (type)  Mode  Cnt    Score    Error  Units
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              2               false                  32  varchar  avgt   50  143.775 ±  1.108  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              2               false                  32   bigint  avgt   50   45.893 ±  0.651  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              2               false                1024  varchar  avgt   50   77.121 ±  0.609  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              2               false                1024   bigint  avgt   50    9.616 ±  0.051  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              2                true                  32  varchar  avgt   50   92.151 ±  0.384  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              2                true                  32   bigint  avgt   50   56.643 ±  0.310  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              2                true                1024  varchar  avgt   50   19.985 ±  0.174  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              2                true                1024   bigint  avgt   50    8.833 ±  0.043  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              4               false                  32  varchar  avgt   50  214.515 ±  0.780  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              4               false                  32   bigint  avgt   50   75.181 ±  1.003  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              4               false                1024  varchar  avgt   50   95.147 ±  0.377  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              4               false                1024   bigint  avgt   50   13.822 ±  0.051  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              4                true                  32  varchar  avgt   50  149.661 ±  0.613  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              4                true                  32   bigint  avgt   50   94.085 ±  0.839  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              4                true                1024  varchar  avgt   50   24.553 ±  0.174  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              4                true                1024   bigint  avgt   50   13.018 ±  0.051  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              8               false                  32   bigint  avgt   50  127.073 ±  1.259  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              8               false                1024  varchar  avgt   50  129.735 ±  0.541  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              8               false                1024   bigint  avgt   50   21.865 ±  0.104  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              8                true                  32  varchar  avgt   50  258.460 ±  0.951  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              8                true                  32   bigint  avgt   50  165.347 ±  2.190  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              8                true                1024  varchar  avgt   50   33.486 ±  0.182  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented              8                true                1024   bigint  avgt   50   21.073 ±  0.075  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             16               false                  32   bigint  avgt   50  221.195 ±  1.120  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             16               false                1024  varchar  avgt   50  199.427 ±  0.476  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             16               false                1024   bigint  avgt   50   37.843 ±  0.175  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             16                true                  32  varchar  avgt   50  485.161 ±  1.569  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             16                true                  32   bigint  avgt   50  295.203 ±  2.633  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             16                true                1024  varchar  avgt   50   52.209 ±  0.265  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             16                true                1024   bigint  avgt   50   37.172 ±  0.220  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             32               false                  32   bigint  avgt   50  413.407 ±  1.834  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             32               false                1024  varchar  avgt   50  339.468 ±  1.265  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             32               false                1024   bigint  avgt   50   69.612 ±  0.218  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             32                true                  32   bigint  avgt   50  584.004 ± 12.342  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             32                true                1024  varchar  avgt   50   89.021 ±  0.305  ms/op
BenchmarkScanFilterAndProjectOperator.benchmarkColumnOriented             32                true                1024   bigint  avgt   50   71.113 ±  0.421  ms/op

sopel39 · 2019-07-08T10:30:52Z

Some queries (significant CPU reduction for TopN queries with filter):

No work processor pipelines:

explain analyze select * from hive.tpch_sf10_orc.lineitem order by partkey limit 10;

 Fragment 2 [SOURCE]
     CPU: 38.48s, Scheduled: 4.65m, Input: 59986052 rows (7.30GB); per task: avg.: 59986052.00 std.dev.: 0.00, Output: 480 rows (71.47kB)
     Output layout: [orderkey, partkey, suppkey, linenumber, quantity, extendedprice, discount, tax, returnflag, linestatus, shipdate, commitdate, receiptdate, shipinstruct, shipmode, comment]
     Output partitioning: SINGLE []
     Stage Execution Strategy: UNGROUPED_EXECUTION
     TopNPartial[10 by (partkey ASC_NULLS_LAST)]
     │   Layout: [orderkey:bigint, partkey:bigint, suppkey:bigint, linenumber:integer, quantity:double, extendedprice:double, discount:double, tax:double, returnflag:varchar(1), linestatus:varchar(1), shipdate:date, commitdate:date, receiptdat
     │   CPU: 4.27s (0.86%), Scheduled: 6.74s (17.52%), Output: 480 rows (71.47kB)
     │   Input avg.: 1153577.92 rows, Input std.dev.: 30.50%
     └─ TableScan[hive:tpch_sf10_orc:lineitem, grouped = false]
            Layout: [orderkey:bigint, partkey:bigint, suppkey:bigint, linenumber:integer, quantity:double, extendedprice:double, discount:double, tax:double, returnflag:varchar(1), linestatus:varchar(1), shipdate:date, commitdate:date, receipt
            Estimates: {rows: ? (?), cpu: ?, memory: 0B, network: 0B}
            CPU: 34.21s (6.86%), Scheduled: 8.20m (1278.45%), Output: 59986052 rows (7.30GB)

presto> explain analyze select * from hive.tpch_sf10_orc.lineitem where partkey < 100000 order by partkey limit 10;

 Fragment 2 [SOURCE]
     CPU: 35.97s, Scheduled: 4.60m, Input: 59986052 rows (7.30GB); per task: avg.: 59986052.00 std.dev.: 0.00, Output: 480 rows (71.47kB)
     Output layout: [orderkey, partkey, suppkey, linenumber, quantity, extendedprice, discount, tax, returnflag, linestatus, shipdate, commitdate, receiptdate, shipinstruct, shipmode, comment]
     Output partitioning: SINGLE []
     Stage Execution Strategy: UNGROUPED_EXECUTION
     TopNPartial[10 by (partkey ASC_NULLS_LAST)]
     │   Layout: [orderkey:bigint, partkey:bigint, suppkey:bigint, linenumber:integer, quantity:double, extendedprice:double, discount:double, tax:double, returnflag:varchar(1), linestatus:varchar(1), shipdate:date, commitdate:date, receiptdat
     │   CPU: 214.00ms (0.04%), Scheduled: 353.00ms (0.98%), Output: 480 rows (71.47kB)
     │   Input avg.: 57638.79 rows, Input std.dev.: 30.50%
     └─ ScanFilter[table = hive:tpch_sf10_orc:lineitem, grouped = false, filterPredicate = ("partkey" < BIGINT '100000')]
            Layout: [orderkey:bigint, partkey:bigint, suppkey:bigint, linenumber:integer, quantity:double, extendedprice:double, discount:double, tax:double, returnflag:varchar(1), linestatus:varchar(1), shipdate:date, commitdate:date, receipt
            Estimates: {rows: ? (?), cpu: ?, memory: 0B, network: 0B}/{rows: ? (?), cpu: ?, memory: 0B, network: 0B}
            CPU: 35.75s (7.40%), Scheduled: 8.05m (1342.71%), Output: 2997217 rows (436.74MB)

with work processor pipelines:

explain analyze select * from hive.tpch_sf10_orc.lineitem order by partkey limit 10;

 Fragment 2 [SOURCE]
     CPU: 20.81s, Scheduled: 3.87m, Input: 59986052 rows (803.03MB); per task: avg.: 59986052.00 std.dev.: 0.00, Output: 480 rows (71.47kB)
     Output layout: [orderkey, partkey, suppkey, linenumber, quantity, extendedprice, discount, tax, returnflag, linestatus, shipdate, commitdate, receiptdate, shipinstruct, shipmode, comment]
     Output partitioning: SINGLE []
     Stage Execution Strategy: UNGROUPED_EXECUTION
     TopNPartial[10 by (partkey ASC_NULLS_LAST)]
     │   Layout: [orderkey:bigint, partkey:bigint, suppkey:bigint, linenumber:integer, quantity:double, extendedprice:double, discount:double, tax:double, returnflag:varchar(1), linestatus:varchar(1), shipdate:date, commitdate:date, receiptdat
     │   CPU: 19.00s (4.37%), Scheduled: 2.90m (836.25%), Output: 480 rows (71.47kB)
     │   Input avg.: 1153577.92 rows, Input std.dev.: 30.50%
     └─ TableScan[hive:tpch_sf10_orc:lineitem, grouped = false]
            Layout: [orderkey:bigint, partkey:bigint, suppkey:bigint, linenumber:integer, quantity:double, extendedprice:double, discount:double, tax:double, returnflag:varchar(1), linestatus:varchar(1), shipdate:date, commitdate:date, receipt
            Estimates: {rows: ? (?), cpu: ?, memory: 0B, network: 0B}
            CPU: 1.80s (0.41%), Scheduled: 4.34m (1252.39%), Output: 59986052 rows (803.03MB)

explain analyze select * from hive.tpch_sf10_orc.lineitem where partkey < 1000000000000 order by partkey limit 10;

 Fragment 2 [SOURCE]
     CPU: 20.64s, Scheduled: 3.88m, Input: 59986052 rows (803.03MB); per task: avg.: 59986052.00 std.dev.: 0.00, Output: 480 rows (71.47kB)
     Output layout: [orderkey, partkey, suppkey, linenumber, quantity, extendedprice, discount, tax, returnflag, linestatus, shipdate, commitdate, receiptdate, shipinstruct, shipmode, comment]
     Output partitioning: SINGLE []
     Stage Execution Strategy: UNGROUPED_EXECUTION
     TopNPartial[10 by (partkey ASC_NULLS_LAST)]
     │   Layout: [orderkey:bigint, partkey:bigint, suppkey:bigint, linenumber:integer, quantity:double, extendedprice:double, discount:double, tax:double, returnflag:varchar(1), linestatus:varchar(1), shipdate:date, commitdate:date, receiptdat
     │   CPU: 16.63s (3.81%), Scheduled: 2.90m (843.18%), Output: 480 rows (71.47kB)
     │   Input avg.: 1153577.92 rows, Input std.dev.: 30.50%
     └─ ScanFilter[table = hive:tpch_sf10_orc:lineitem, grouped = false, filterPredicate = ("partkey" < 1000000000000)]
            Layout: [orderkey:bigint, partkey:bigint, suppkey:bigint, linenumber:integer, quantity:double, extendedprice:double, discount:double, tax:double, returnflag:varchar(1), linestatus:varchar(1), shipdate:date, commitdate:date, receipt
            Estimates: {rows: ? (?), cpu: ?, memory: 0B, network: 0B}/{rows: ? (?), cpu: ?, memory: 0B, network: 0B}
            CPU: 4.00s (0.92%), Scheduled: 4.37m (1270.81%), Output: 59986052 rows (803.03MB)

presto> explain analyze select * from hive.tpch_sf10_orc.lineitem where partkey < 100000 order by partkey limit 10;

 Fragment 2 [SOURCE]
     CPU: 16.59s, Scheduled: 3.70m, Input: 59986052 rows (802.87MB); per task: avg.: 59986052.00 std.dev.: 0.00, Output: 480 rows (71.47kB)
     Output layout: [orderkey, partkey, suppkey, linenumber, quantity, extendedprice, discount, tax, returnflag, linestatus, shipdate, commitdate, receiptdate, shipinstruct, shipmode, comment]
     Output partitioning: SINGLE []
     Stage Execution Strategy: UNGROUPED_EXECUTION
     TopNPartial[10 by (partkey ASC_NULLS_LAST)]
     │   Layout: [orderkey:bigint, partkey:bigint, suppkey:bigint, linenumber:integer, quantity:double, extendedprice:double, discount:double, tax:double, returnflag:varchar(1), linestatus:varchar(1), shipdate:date, commitdate:date, receiptdat
     │   CPU: 12.80s (3.04%), Scheduled: 2.73m (987.65%), Output: 480 rows (71.47kB)
     │   Input avg.: 57638.79 rows, Input std.dev.: 30.50%
     └─ ScanFilter[table = hive:tpch_sf10_orc:lineitem, grouped = false, filterPredicate = ("partkey" < BIGINT '100000')]
            Layout: [orderkey:bigint, partkey:bigint, suppkey:bigint, linenumber:integer, quantity:double, extendedprice:double, discount:double, tax:double, returnflag:varchar(1), linestatus:varchar(1), shipdate:date, commitdate:date, receipt
            Estimates: {rows: ? (?), cpu: ?, memory: 0B, network: 0B}/{rows: ? (?), cpu: ?, memory: 0B, network: 0B}
            CPU: 3.78s (0.90%), Scheduled: 4.30m (1554.56%), Output: 2997217 rows (40.70MB)

sopel39 · 2019-07-17T09:42:23Z

part of: #49

dain

Mostly just suggestions, and requests for more explanations (and code comments in one case).

I'm curious why we are only applying lazyness to some cases, how those were choose, and if we should expose it to more cases?

presto-main/src/main/java/io/prestosql/operator/project/InputPageProjection.java

presto-main/src/main/java/io/prestosql/operator/project/DictionaryAwarePageProjection.java

presto-main/src/main/java/io/prestosql/operator/project/PageProcessor.java

presto-main/src/main/java/io/prestosql/operator/project/MergePages.java

presto-main/src/main/java/io/prestosql/operator/PageUtils.java

presto-main/src/main/java/io/prestosql/operator/project/PageProcessor.java

sopel39

ac

presto-main/src/main/java/io/prestosql/operator/PageUtils.java

presto-main/src/main/java/io/prestosql/operator/project/PageProcessor.java

sopel39 added the WIP label Jul 3, 2019

cla-bot bot added the cla-signed label Jul 3, 2019

sopel39 force-pushed the ks/lazy_page_processor branch 2 times, most recently from fa91b33 to f49bddb Compare July 4, 2019 16:19

Add WorkProcessorSourceOperatorFactory#getPlanNodeId method

7e694b6

ScanFilterAndProject can have different source node ID and plan node ID.

sopel39 force-pushed the ks/lazy_page_processor branch 5 times, most recently from 038b4f4 to 2cd3415 Compare July 6, 2019 22:21

sopel39 removed the WIP label Jul 8, 2019

sopel39 changed the title ~~[WIP] Make PageProcessor preserve input block laziness~~ Make PageProcessor preserve input block laziness Jul 8, 2019

sopel39 requested a review from dain July 8, 2019 10:30

sopel39 assigned dain Jul 17, 2019

Make WorkProcessorSourceOperator adapter be aware of creation context

ef4e0b8

sopel39 force-pushed the ks/lazy_page_processor branch from 2cd3415 to becb49f Compare July 23, 2019 10:01

dain requested changes Jul 25, 2019

View reviewed changes

sopel39 force-pushed the ks/lazy_page_processor branch from becb49f to c97a15b Compare July 26, 2019 08:18

sopel39 commented Jul 26, 2019

View reviewed changes

presto-main/src/main/java/io/prestosql/operator/PageUtils.java Show resolved Hide resolved

presto-main/src/main/java/io/prestosql/operator/PageUtils.java Show resolved Hide resolved

presto-main/src/main/java/io/prestosql/operator/project/PageProcessor.java Show resolved Hide resolved

sopel39 added 2 commits July 26, 2019 19:46

Make InputPageProjection produce lazy pages

c03c3a5

Add support for lazy blocks to DictionaryAwarePageProjection

e08f965

sopel39 force-pushed the ks/lazy_page_processor branch from c97a15b to 3278c62 Compare July 26, 2019 17:48

sopel39 added 4 commits July 26, 2019 20:04

Avoid page materialization in PageProcessor

274a15e

Skip merge operator for lazy blocks in work processor

8baf88a

Call sizeInBytes consumer once for loaded blocks

0552729

Do not instantiate new Page when all Blocks are non lazy

ba52e16

sopel39 force-pushed the ks/lazy_page_processor branch from 3278c62 to ba52e16 Compare July 26, 2019 18:04

dain self-requested a review July 27, 2019 21:25

dain approved these changes Jul 28, 2019

View reviewed changes

dain assigned sopel39 and unassigned dain Jul 28, 2019

sopel39 merged commit 25da929 into trinodb:master Jul 29, 2019

sopel39 deleted the ks/lazy_page_processor branch July 29, 2019 08:41

sopel39 mentioned this pull request Jul 29, 2019

Release notes for 317 #1098

Closed

5 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Make PageProcessor preserve input block laziness #1075

Make PageProcessor preserve input block laziness #1075

sopel39 commented Jul 3, 2019 •

edited

Loading

sopel39 commented Jul 8, 2019 •

edited

Loading

sopel39 commented Jul 8, 2019 •

edited

Loading

sopel39 commented Jul 17, 2019

dain left a comment

sopel39 left a comment

Make PageProcessor preserve input block laziness #1075

Make PageProcessor preserve input block laziness #1075

Conversation

sopel39 commented Jul 3, 2019 • edited Loading

sopel39 commented Jul 8, 2019 • edited Loading

sopel39 commented Jul 8, 2019 • edited Loading

sopel39 commented Jul 17, 2019

dain left a comment

Choose a reason for hiding this comment

sopel39 left a comment

Choose a reason for hiding this comment

sopel39 commented Jul 3, 2019 •

edited

Loading

sopel39 commented Jul 8, 2019 •

edited

Loading

sopel39 commented Jul 8, 2019 •

edited

Loading