[jira] [Resolved] (DRILL-4577) Improve performance for query on INFORMATION_SCHEMA when HIVE is plugged in
[ https://issues.apache.org/jira/browse/DRILL-4577?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Sean Hsuan-Yi Chu resolved DRILL-4577. -- Resolution: Fixed Resolved in commit-id: b8f6ebc651445ccecd3e393250f6cd2781fc07e3 > Improve performance for query on INFORMATION_SCHEMA when HIVE is plugged in > --- > > Key: DRILL-4577 > URL: https://issues.apache.org/jira/browse/DRILL-4577 > Project: Apache Drill > Issue Type: Improvement > Components: Storage - Hive >Reporter: Sean Hsuan-Yi Chu >Assignee: Sean Hsuan-Yi Chu > Fix For: 1.7.0 > > > A query such as > {code} > select * from INFORMATION_SCHEMA.`TABLES` > {code} > is converted as calls to fetch all tables from storage plugins. > When users have Hive, the calls to hive metadata storage would be: > 1) get_table > 2) get_partitions > However, the information regarding partitions is not used in this type of > queries. Beside, a more efficient way is to fetch tables is to use > get_multi_table call. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (DRILL-4577) Improve performance for query on INFORMATION_SCHEMA when HIVE is plugged in
[ https://issues.apache.org/jira/browse/DRILL-4577?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15268148#comment-15268148 ] ASF GitHub Bot commented on DRILL-4577: --- Github user asfgit closed the pull request at: https://github.com/apache/drill/pull/461 > Improve performance for query on INFORMATION_SCHEMA when HIVE is plugged in > --- > > Key: DRILL-4577 > URL: https://issues.apache.org/jira/browse/DRILL-4577 > Project: Apache Drill > Issue Type: Improvement > Components: Storage - Hive >Reporter: Sean Hsuan-Yi Chu >Assignee: Sean Hsuan-Yi Chu > Fix For: 1.7.0 > > > A query such as > {code} > select * from INFORMATION_SCHEMA.`TABLES` > {code} > is converted as calls to fetch all tables from storage plugins. > When users have Hive, the calls to hive metadata storage would be: > 1) get_table > 2) get_partitions > However, the information regarding partitions is not used in this type of > queries. Beside, a more efficient way is to fetch tables is to use > get_multi_table call. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (DRILL-4577) Improve performance for query on INFORMATION_SCHEMA when HIVE is plugged in
[ https://issues.apache.org/jira/browse/DRILL-4577?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15267771#comment-15267771 ] ASF GitHub Bot commented on DRILL-4577: --- Github user jinfengni commented on the pull request: https://github.com/apache/drill/pull/461#issuecomment-216399417 +1. The patch looks good to me. Internally performance measurement shows orders of magnitude improvement for hive schema with up to 32k tables. > Improve performance for query on INFORMATION_SCHEMA when HIVE is plugged in > --- > > Key: DRILL-4577 > URL: https://issues.apache.org/jira/browse/DRILL-4577 > Project: Apache Drill > Issue Type: Improvement > Components: Storage - Hive >Reporter: Sean Hsuan-Yi Chu >Assignee: Sean Hsuan-Yi Chu > Fix For: 1.7.0 > > > A query such as > {code} > select * from INFORMATION_SCHEMA.`TABLES` > {code} > is converted as calls to fetch all tables from storage plugins. > When users have Hive, the calls to hive metadata storage would be: > 1) get_table > 2) get_partitions > However, the information regarding partitions is not used in this type of > queries. Beside, a more efficient way is to fetch tables is to use > get_multi_table call. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Comment Edited] (DRILL-4577) Improve performance for query on INFORMATION_SCHEMA when HIVE is plugged in
[ https://issues.apache.org/jira/browse/DRILL-4577?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15267726#comment-15267726 ] Sean Hsuan-Yi Chu edited comment on DRILL-4577 at 5/2/16 11:34 PM: --- [~vkorukanti], Data points regarding Performance: 1. 1k tables in the DB: Without this patch, it took around 550 seconds; With this patch, it took about 3.7 - 4.3 seconds; 2. 32k tables in the DB: Without this patch, the result does not come back. With this patch, it took about 83.1 seconds; It seems to me that the current performance is not acceptable when #of tables is beyond 1k. With the points here, I think this patch, along with the option, is really needed. was (Author: seanhychu): [~vkorukanti], Data points regarding Performance: 1. 1k tables in the DB: Without this patch, it took around 550 seconds; With this patch, it took about 3.7 - 4.3 seconds; 2. 32k tables in the DB: Without this patch, the result does not come back. With this patch, it took about 83.1 seconds; With the points here, I think this patch, along with the option, is really needed. > Improve performance for query on INFORMATION_SCHEMA when HIVE is plugged in > --- > > Key: DRILL-4577 > URL: https://issues.apache.org/jira/browse/DRILL-4577 > Project: Apache Drill > Issue Type: Improvement > Components: Storage - Hive >Reporter: Sean Hsuan-Yi Chu >Assignee: Sean Hsuan-Yi Chu > Fix For: 1.7.0 > > > A query such as > {code} > select * from INFORMATION_SCHEMA.`TABLES` > {code} > is converted as calls to fetch all tables from storage plugins. > When users have Hive, the calls to hive metadata storage would be: > 1) get_table > 2) get_partitions > However, the information regarding partitions is not used in this type of > queries. Beside, a more efficient way is to fetch tables is to use > get_multi_table call. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (DRILL-4577) Improve performance for query on INFORMATION_SCHEMA when HIVE is plugged in
[ https://issues.apache.org/jira/browse/DRILL-4577?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15267726#comment-15267726 ] Sean Hsuan-Yi Chu commented on DRILL-4577: -- [~vkorukanti], Data points regarding Performance: 1. 1k tables in the DB: Without this patch, it took around 550 seconds; With this patch, it took about 3.7 - 4.3 seconds; 2. 32k tables in the DB: Without this patch, the result does not come back. With this patch, it took about 83.1 seconds; With the points here, I think this patch, along with the option, is really needed. > Improve performance for query on INFORMATION_SCHEMA when HIVE is plugged in > --- > > Key: DRILL-4577 > URL: https://issues.apache.org/jira/browse/DRILL-4577 > Project: Apache Drill > Issue Type: Improvement > Components: Storage - Hive >Reporter: Sean Hsuan-Yi Chu >Assignee: Sean Hsuan-Yi Chu > Fix For: 1.7.0 > > > A query such as > {code} > select * from INFORMATION_SCHEMA.`TABLES` > {code} > is converted as calls to fetch all tables from storage plugins. > When users have Hive, the calls to hive metadata storage would be: > 1) get_table > 2) get_partitions > However, the information regarding partitions is not used in this type of > queries. Beside, a more efficient way is to fetch tables is to use > get_multi_table call. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (DRILL-4641) Support for lzo compression
[ https://issues.apache.org/jira/browse/DRILL-4641?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15267595#comment-15267595 ] subbu srinivasan commented on DRILL-4641: - Jason, Need to make following config changes to make it working. - Modify core-site.xml to include the following. The property specifies the list of codecs that will be exposed by the compression interface (org.apache.hadoop.io.compress.CompressionCodecFactory and org.apache.hadoop.io.compress.CompressionCodec) io.compression.codecs org.apache.hadoop.io.compress.DefaultCodec,org.apache.hadoop.io.compress.GzipCodec, org.apache.hadoop.io.compress.BZip2Codec,com.hadoop.compression.lzo.LzopCodec - Download the lzo java compression files - lzo-hadoop-1.0.5.jar and lzo-core-1.0.5.jar - Define the extension appropriately in the storage plugin "json": { "type": "json", "extensions": [ "lzo" ] }, This got me going. > Support for lzo compression > --- > > Key: DRILL-4641 > URL: https://issues.apache.org/jira/browse/DRILL-4641 > Project: Apache Drill > Issue Type: Improvement > Components: Storage - Other >Affects Versions: Future > Environment: Not specific to platform >Reporter: subbu srinivasan > > Would love support for quering lzo compressed files. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Closed] (DRILL-4472) Pushing Filter past Union All fails: DRILL-3257 regressed DRILL-2746 but unit test update break test goal
[ https://issues.apache.org/jira/browse/DRILL-4472?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Krystal closed DRILL-4472. -- Per Hsuan, this jira is to rewrite a unit test so there is nothing to verify. > Pushing Filter past Union All fails: DRILL-3257 regressed DRILL-2746 but unit > test update break test goal > - > > Key: DRILL-4472 > URL: https://issues.apache.org/jira/browse/DRILL-4472 > Project: Apache Drill > Issue Type: Bug > Components: Query Planning & Optimization >Reporter: Jacques Nadeau >Assignee: Sean Hsuan-Yi Chu > Fix For: 1.7.0 > > > While reviewing DRILL-4467, I discovered this test. > https://github.com/apache/drill/blame/master/exec/java-exec/src/test/java/org/apache/drill/TestUnionAll.java#L560 > As you can see, the test is checking that test name confirms that filter is > pushed below union all. However, as you can see, the expected result in > DRILL-3257 was updated to a plan which doesn't push the in clause below the > filter. I'm disabling the test since 4467 happens to remove what becomes a > trivial project. However, we really should fix the core problem (a regression > of DRILL-2746. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Closed] (DRILL-3745) Hive CHAR not supported
[ https://issues.apache.org/jira/browse/DRILL-3745?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Krystal closed DRILL-3745. -- git.commit.id.abbrev=5705d45 Able to run queries against hive tables with char data type. > Hive CHAR not supported > --- > > Key: DRILL-3745 > URL: https://issues.apache.org/jira/browse/DRILL-3745 > Project: Apache Drill > Issue Type: Bug >Affects Versions: 1.1.0 >Reporter: Nathaniel Auvil >Assignee: Arina Ielchiieva > Labels: doc-impacting > Fix For: 1.7.0 > > > It doesn’t look like Drill 1.1.0 supports the Hive CHAR type? > In Hive: > create table development.foo > ( > bad CHAR(10) > ); > And then in sqlline: > > use `hive.development`; > > select * from foo; > Error: PARSE ERROR: Unsupported Hive data type CHAR. > Following Hive data types are supported in Drill INFORMATION_SCHEMA: > BOOLEAN, BYTE, SHORT, INT, LONG, FLOAT, DOUBLE, DATE, TIMESTAMP, > BINARY, DECIMAL, STRING, VARCHAR, LIST, MAP, STRUCT and UNION > [Error Id: 58bf3940-3c09-4ad2-8f52-d052dffd4b17 on dtpg05:31010] > (state=,code=0) > This was originally found when getting failures trying to connect via JDBS > using Squirrel. We have the Hive plugin enabled with tables using CHAR. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Closed] (DRILL-4529) SUM() with windows function result in mismatch nullability
[ https://issues.apache.org/jira/browse/DRILL-4529?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Krystal closed DRILL-4529. -- git.commit.id.abbrev=5705d45 Verified that bug is fixed. > SUM() with windows function result in mismatch nullability > -- > > Key: DRILL-4529 > URL: https://issues.apache.org/jira/browse/DRILL-4529 > Project: Apache Drill > Issue Type: Bug > Components: Query Planning & Optimization >Reporter: Krystal >Assignee: Sean Hsuan-Yi Chu > Labels: limit0 > Fix For: 1.7.0 > > > git.commit.id.abbrev=cee5317 > select > sum(1) over w sum1, > sum(5) over w sum5, > sum(10) over w sum10 > from > j1_v > where > c_date is not null > window w as (partition by c_date); > Output from test: > limit 0: [columnNoNulls, columnNoNulls, columnNoNulls] > regular: [columnNullable, columnNullable, columnNullable] -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (DRILL-3894) Directory functions (MaxDir, MinDir ..) should have optional filename parameter
[ https://issues.apache.org/jira/browse/DRILL-3894?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15267069#comment-15267069 ] ASF GitHub Bot commented on DRILL-3894: --- Github user parthchandra commented on the pull request: https://github.com/apache/drill/pull/467#issuecomment-216307896 Looks good. +1 > Directory functions (MaxDir, MinDir ..) should have optional filename > parameter > --- > > Key: DRILL-3894 > URL: https://issues.apache.org/jira/browse/DRILL-3894 > Project: Apache Drill > Issue Type: Improvement > Components: Functions - Drill >Affects Versions: 1.2.0 >Reporter: Neeraja >Assignee: Vitalii Diravka > > https://drill.apache.org/docs/query-directory-functions/ > The directory functions documented above should provide ability to have > second parameter(file name) as optional. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (DRILL-4573) Zero copy LIKE, REGEXP_MATCHES, SUBSTR
[ https://issues.apache.org/jira/browse/DRILL-4573?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15266840#comment-15266840 ] Jacques Nadeau commented on DRILL-4573: --- When we started working on functions, immutability wasn't a constraint so we may have a few functions around that don't follow this rule. The goal is go back and correct those but make sure we don't add any new mutations. > Zero copy LIKE, REGEXP_MATCHES, SUBSTR > -- > > Key: DRILL-4573 > URL: https://issues.apache.org/jira/browse/DRILL-4573 > Project: Apache Drill > Issue Type: Improvement >Reporter: jean-claude >Priority: Minor > Fix For: 1.7.0 > > Attachments: DRILL-4573-3.patch.txt, DRILL-4573.patch.txt > > > All the functions using the java.util.regex.Matcher are currently creating > Java string objects to pass into the matcher.reset(). > However this creates unnecessary copy of the bytes and a Java string object. > The matcher uses a CharSequence, so instead of making a copy we can create an > adapter from the DrillBuffer to the CharSequence interface. > Gains of 25% in execution speed are possible when going over VARCHAR of 36 > chars. The gain will be proportional to the size of the VARCHAR. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Assigned] (DRILL-3149) TextReader should support multibyte line delimiters
[ https://issues.apache.org/jira/browse/DRILL-3149?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Arina Ielchiieva reassigned DRILL-3149: --- Assignee: Arina Ielchiieva > TextReader should support multibyte line delimiters > --- > > Key: DRILL-3149 > URL: https://issues.apache.org/jira/browse/DRILL-3149 > Project: Apache Drill > Issue Type: Improvement > Components: Storage - Text & CSV >Affects Versions: 1.0.0, 1.1.0 >Reporter: Jim Scott >Assignee: Arina Ielchiieva >Priority: Minor > Fix For: Future > > > lineDelimiter in the TextFormatConfig doesn't support \r\n for record > delimiters. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (DRILL-3474) Add implicit file columns support
[ https://issues.apache.org/jira/browse/DRILL-3474?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15266717#comment-15266717 ] ASF GitHub Bot commented on DRILL-3474: --- GitHub user arina-ielchiieva opened a pull request: https://github.com/apache/drill/pull/491 DRILL-3474: Add implicit file columns support You can merge this pull request into a Git repository by running: $ git pull https://github.com/arina-ielchiieva/drill DRILL-3474 Alternatively you can review and apply these changes as the patch at: https://github.com/apache/drill/pull/491.patch To close this pull request, make a commit to your master/trunk branch with (at least) the following in the commit message: This closes #491 commit aff44f3fafa35d946115899f9e9941e2b2af22d6 Author: Arina IelchiievaDate: 2016-04-18T16:36:52Z DRILL-3474: Add implicit file columns support > Add implicit file columns support > - > > Key: DRILL-3474 > URL: https://issues.apache.org/jira/browse/DRILL-3474 > Project: Apache Drill > Issue Type: Bug > Components: Metadata >Affects Versions: 1.1.0 >Reporter: Jim Scott >Assignee: Arina Ielchiieva > Fix For: Future > > > I could not find another ticket which talks about this ... > The file name should be a column which can be selected or filtered when > querying a directory just like dir0, dir1 are available. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (DRILL-4573) Zero copy LIKE, REGEXP_MATCHES, SUBSTR
[ https://issues.apache.org/jira/browse/DRILL-4573?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15266546#comment-15266546 ] jean-claude commented on DRILL-4573: ok, I've made the change. > Zero copy LIKE, REGEXP_MATCHES, SUBSTR > -- > > Key: DRILL-4573 > URL: https://issues.apache.org/jira/browse/DRILL-4573 > Project: Apache Drill > Issue Type: Improvement >Reporter: jean-claude >Priority: Minor > Fix For: 1.7.0 > > Attachments: DRILL-4573-3.patch.txt, DRILL-4573.patch.txt > > > All the functions using the java.util.regex.Matcher are currently creating > Java string objects to pass into the matcher.reset(). > However this creates unnecessary copy of the bytes and a Java string object. > The matcher uses a CharSequence, so instead of making a copy we can create an > adapter from the DrillBuffer to the CharSequence interface. > Gains of 25% in execution speed are possible when going over VARCHAR of 36 > chars. The gain will be proportional to the size of the VARCHAR. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (DRILL-4573) Zero copy LIKE, REGEXP_MATCHES, SUBSTR
[ https://issues.apache.org/jira/browse/DRILL-4573?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] jean-claude updated DRILL-4573: --- Attachment: DRILL-4573-3.patch.txt > Zero copy LIKE, REGEXP_MATCHES, SUBSTR > -- > > Key: DRILL-4573 > URL: https://issues.apache.org/jira/browse/DRILL-4573 > Project: Apache Drill > Issue Type: Improvement >Reporter: jean-claude >Priority: Minor > Fix For: 1.7.0 > > Attachments: DRILL-4573-3.patch.txt, DRILL-4573.patch.txt > > > All the functions using the java.util.regex.Matcher are currently creating > Java string objects to pass into the matcher.reset(). > However this creates unnecessary copy of the bytes and a Java string object. > The matcher uses a CharSequence, so instead of making a copy we can create an > adapter from the DrillBuffer to the CharSequence interface. > Gains of 25% in execution speed are possible when going over VARCHAR of 36 > chars. The gain will be proportional to the size of the VARCHAR. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (DRILL-4573) Zero copy LIKE, REGEXP_MATCHES, SUBSTR
[ https://issues.apache.org/jira/browse/DRILL-4573?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] jean-claude updated DRILL-4573: --- Attachment: (was: DRILL-4573-3.patch.txt) > Zero copy LIKE, REGEXP_MATCHES, SUBSTR > -- > > Key: DRILL-4573 > URL: https://issues.apache.org/jira/browse/DRILL-4573 > Project: Apache Drill > Issue Type: Improvement >Reporter: jean-claude >Priority: Minor > Fix For: 1.7.0 > > Attachments: DRILL-4573-3.patch.txt, DRILL-4573.patch.txt > > > All the functions using the java.util.regex.Matcher are currently creating > Java string objects to pass into the matcher.reset(). > However this creates unnecessary copy of the bytes and a Java string object. > The matcher uses a CharSequence, so instead of making a copy we can create an > adapter from the DrillBuffer to the CharSequence interface. > Gains of 25% in execution speed are possible when going over VARCHAR of 36 > chars. The gain will be proportional to the size of the VARCHAR. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (DRILL-4573) Zero copy LIKE, REGEXP_MATCHES, SUBSTR
[ https://issues.apache.org/jira/browse/DRILL-4573?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15266437#comment-15266437 ] jean-claude commented on DRILL-4573: ok. I had it as a copy then saw the REPLACE function doing out.buffer = text.buffer, but now I see that text is a constant buffer. I'll change it back. > Zero copy LIKE, REGEXP_MATCHES, SUBSTR > -- > > Key: DRILL-4573 > URL: https://issues.apache.org/jira/browse/DRILL-4573 > Project: Apache Drill > Issue Type: Improvement >Reporter: jean-claude >Priority: Minor > Fix For: 1.7.0 > > Attachments: DRILL-4573-3.patch.txt, DRILL-4573.patch.txt > > > All the functions using the java.util.regex.Matcher are currently creating > Java string objects to pass into the matcher.reset(). > However this creates unnecessary copy of the bytes and a Java string object. > The matcher uses a CharSequence, so instead of making a copy we can create an > adapter from the DrillBuffer to the CharSequence interface. > Gains of 25% in execution speed are possible when going over VARCHAR of 36 > chars. The gain will be proportional to the size of the VARCHAR. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (DRILL-3878) Support XML Querying (selects/projections, no writing)
[ https://issues.apache.org/jira/browse/DRILL-3878?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15266184#comment-15266184 ] ASF GitHub Bot commented on DRILL-3878: --- Github user magpierre closed the pull request at: https://github.com/apache/drill/pull/451 > Support XML Querying (selects/projections, no writing) > -- > > Key: DRILL-3878 > URL: https://issues.apache.org/jira/browse/DRILL-3878 > Project: Apache Drill > Issue Type: New Feature >Affects Versions: Future >Reporter: Edmon Begoli > Labels: features > Fix For: 1.7.0 > > Original Estimate: 3,360h > Remaining Estimate: 3,360h > > Support querying of the XML documents (as read-only selects, > Writing should be implemented as a different feature that brings its own set > of challenges.) > To consider is reading of the trivial, schema-less, XML documents, > DTD-oriented ones and also of schema-defined ones. > Also, we should consider direct querying vs. using converter tools to change > the representation from XML to JSON, CSV, etc. > Design and Implementation discussion, notes, ideas and implementation > suggestions should be captured here: > https://docs.google.com/document/d/1oS-cObSaTlAmuW_XghDLmHbBEorLl0z-axaHnjy7vg0/edit?usp=sharing > > (no vandalism, please) -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (DRILL-3806) add metadata for untyped null and simple type promotion
[ https://issues.apache.org/jira/browse/DRILL-3806?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15266171#comment-15266171 ] Joris Gillis commented on DRILL-3806: - I am experiencing the same issue. Any progress in the meantime? > add metadata for untyped null and simple type promotion > --- > > Key: DRILL-3806 > URL: https://issues.apache.org/jira/browse/DRILL-3806 > Project: Apache Drill > Issue Type: Bug > Components: Execution - Data Types >Reporter: Julien Le Dem > Fix For: Future > > > Currently when a field has literal null values in JSON the type will be > assigned as BIGINT by default for lack of better type. > ``` > { > "a": null > } > ``` > if later on a is assigned with a string value the query will fail with a > schema change error, > The idea is to capture the notion of "untyped null" and implement simple type > promotion from untyped null to the actual type. -- This message was sent by Atlassian JIRA (v6.3.4#6332)