Merge from master #1
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
If the current and new values are the same do not set jmx export otherwise an exception is thrown.
Expose elements in the queue via an iterator.
Added 2 new properties to the resource groups: queuedTimeLimit and runningTimeLimit.
Allow repeating an element for a given number of times. The return result is an array.
Add array aggregation benchmark to measure the performance for double-typed array aggregation. Extend the ablility of AbstractOperatorBenchmark to expose peak memory usage.
Make BlockBuilderStatus nullable in various block builders to reduce the memory usage for functions that aggregate columns into complex types; e.g., array_agg, map_agg, histogram, min_by, etc. In theory, the change can reduce memory utilization up to 43%, which is the ratio of the size of BlockBuilderStatus over the size of BlockBuilder. In practice, we may observe a memory safe around 25 - 30% for large scale of data. Small-scale benchmark for array_agg on double type: before: sql_double_array_agg :: 33.493 cpu ms :: 3.13MB peak memory after: sql_double_array_agg :: 33.414 cpu ms :: 2.33MB peak memory Large-scale benchmark for array_agg on double type: rows memory (before) memory (after) save 20 M 4.79 GB 3.50 GB 27% 30 M 7.53 GB 5.64 GB 25% 40 M 9.86 GB 7.03 GB 28%
The thee last test cases in TestTransformExistsApplyToScalarApply#testDoesNotFire do not differ much and there is not much sense to keep them all. This commit removes two of them.
LateralJoin is preferred to be used in case of scalar subqueries.
Returning only input properties for apply node is incorrect as it does not consider subquery properties.
AddLocalExcanges is using PropertyDerivation classes which do not support subquery related plan nodes like ApplyNode.
nonReserved cannot contain recursive rules.
The mechanism for handling non-reserved keywords relies on being able to intercept the parsing operation and replacing the nonReserved entry in the parse tree with an IDENT token. Modifying the tree in this manner breaks when dealing with intermediate nodes, presumably due to how enter/exit rule notification is weaved into the traversal/building of the tree.
We need to allow '.' in resource group name for backward compatibility, so '.' cannot be used as delimiter to parse '.' separated segmented name.
Make Util.pruneInputs return Optional<Set<Symbol>>, rather than Optional<List<Symbol>>, because for most nodes we're thinking about sets of used symbols, rather than the ordering of the output symbol list. In two specific cases, ValueNode and TableScanNode, retain the previous ordering explicitly, rather than losing it in the set, to avoid making superfluous changes to the node during optimization. The PruneJoinColumns rule was already retaining the order of its outputs, and this patch does not affect that behavior.
The Travis builds hang on TestHiveTableStatistics with the new images. https://blog.travis-ci.com/2017-06-21-trusty-updates-2017-Q2-launch
If page.getSizeInBytes() is zero, the page may not be returned to the client. The only known case when this happens is if the page contains only SliceArrayBlocks and all values are null.
After executing a schema-altering query such as CREATE TABLE, datastax driver waits for a certain time period for "schema agreement", before refreshing the schema. The default timeout is 10 seconds, which might be low for schema changes to be propagated to all the nodes.
to account for the fact that it may take some time for the schema metadata to be refreshed. This is similar to the retry mechanism that was added to assertContains() method. Hopefully this will fix the intermittent test failures.
The Hive connector lists partition names during planning, then fetches metadata separately while generating splits. The latter can fail if the partition is dropped in between the two metastore calls.
Detect the MySql error ER_TRANS_CACHE_FULL and generate an additional error message.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.