[jira] [Commented] (DRILL-6937) sys.functions table needs a fix in the names column
[
https://issues.apache.org/jira/browse/DRILL-6937?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16748442#comment-16748442
]
Kunal Khatua commented on DRILL-6937:
-
I think, everything is implemented as a function. So in this case, the function
is
{{bit Z = greater_than(bigint X, bigint Y)}}
However, the SQL language parser is (probably) providing us syntactic sugar by
making it easier to right
{{Z = X > Y}}
Both options are available to the user, we just tend to use what is more
intuitive.
{code}
0: jdbc:drill:drillbit=kk127> select 2 > 1 from (values(1));
+-+
| EXPR$0 |
+-+
| true|
+-+
1 row selected (0.862 seconds)
0: jdbc:drill:drillbit=kk127> select greater_than(2,1) from (values(1));
+-+
| EXPR$0 |
+-+
| true|
+-+
1 row selected (0.453 seconds)
{code}
> sys.functions table needs a fix in the names column
> ---
>
> Key: DRILL-6937
> URL: https://issues.apache.org/jira/browse/DRILL-6937
> Project: Apache Drill
> Issue Type: Bug
> Components: Execution - Flow
>Affects Versions: 1.15.0
>Reporter: Khurram Faraaz
>Assignee: Kunal Khatua
>Priority: Minor
> Fix For: 1.16.0
>
>
> The function names in the name column of sys.functions in some cases, are the
> operators, this is not the expected behavior, the name column should have
> actual names and not the operators.
> I am on Drill 1.15.0 commit : 8743e8f1e8d5bca4d67c94d07a8560ad356ff2b6
> {noformat}
> Apache Drill 1.15.0
> "Data is the new oil. Ready to Drill some?"
> 0: jdbc:drill:schema=dfs.tmp> select count(*) from sys.functions;
> +-+
> | EXPR$0 |
> +-+
> | 2846 |
> +-+
> 1 row selected (0.327 seconds)
> 0: jdbc:drill:schema=dfs.tmp>
> {noformat}
> {noformat}
> 0: jdbc:drill:schema=dfs.tmp> select distinct name from sys.functions limit
> 12;
> ++
> | name |
> ++
> | != |
> | $sum0 |
> | && |
> | - |
> | /int |
> | < |
> | <= |
> | <> |
> | = |
> | == |
> | > |
> | >= |
> ++
> 12 rows selected (0.175 seconds)
> {noformat}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Commented] (DRILL-6937) sys.functions table needs a fix in the names column
[
https://issues.apache.org/jira/browse/DRILL-6937?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16748292#comment-16748292
]
Khurram Faraaz commented on DRILL-6937:
---
One question is, why do we have two different entries in the sys.functions
table, one with name being ">" and other with name being "greater_than", all
other columns for the two functions seem to have identical values, except for
the name.
{noformat}
| > | BIGINT-REQUIRED,BIGINT-REQUIRED | BIT | built-in | false |
| greater_than | BIGINT-REQUIRED,BIGINT-REQUIRED | BIT | built-in | false |
{noformat}
> sys.functions table needs a fix in the names column
> ---
>
> Key: DRILL-6937
> URL: https://issues.apache.org/jira/browse/DRILL-6937
> Project: Apache Drill
> Issue Type: Bug
> Components: Execution - Flow
>Affects Versions: 1.15.0
>Reporter: Khurram Faraaz
>Assignee: Kunal Khatua
>Priority: Minor
> Fix For: 1.16.0
>
>
> The function names in the name column of sys.functions in some cases, are the
> operators, this is not the expected behavior, the name column should have
> actual names and not the operators.
> I am on Drill 1.15.0 commit : 8743e8f1e8d5bca4d67c94d07a8560ad356ff2b6
> {noformat}
> Apache Drill 1.15.0
> "Data is the new oil. Ready to Drill some?"
> 0: jdbc:drill:schema=dfs.tmp> select count(*) from sys.functions;
> +-+
> | EXPR$0 |
> +-+
> | 2846 |
> +-+
> 1 row selected (0.327 seconds)
> 0: jdbc:drill:schema=dfs.tmp>
> {noformat}
> {noformat}
> 0: jdbc:drill:schema=dfs.tmp> select distinct name from sys.functions limit
> 12;
> ++
> | name |
> ++
> | != |
> | $sum0 |
> | && |
> | - |
> | /int |
> | < |
> | <= |
> | <> |
> | = |
> | == |
> | > |
> | >= |
> ++
> 12 rows selected (0.175 seconds)
> {noformat}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Commented] (DRILL-6937) sys.functions table needs a fix in the names column
[
https://issues.apache.org/jira/browse/DRILL-6937?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16748280#comment-16748280
]
Kunal Khatua commented on DRILL-6937:
-
[~khfaraaz] the {{/int}} function name is correct. See
https://github.com/apache/drill/blame/master/exec/java-exec/src/main/codegen/templates/DateIntervalFunctionTemplates/IntervalNumericArithmetic.java#L133
That said, I'm not sure if this is really a bug, since there have been other
functions, like {{$sum0}} around as well:
https://github.com/apache/drill/blame/master/exec/java-exec/src/main/codegen/templates/SumZeroAggr.java#L48
Not all functions in the {{sys.functions}} table are necessarily utilized by
users in their alphabetical representation, and are exposed to mathematical
symbols as operators.
For example, the _less than_ or _greater than_ operator can be seen here in
multiple forms.
{code}
0: jdbc:drill:drillbit=kk127> select * from sys.functions where returnType =
'BIT' and signature = 'BIGINT-REQUIRED,BIGINT-REQUIRED' limit 15;
+---+--+-+---+---+
| name|signature | returnType |
source | internal |
+---+--+-+---+---+
| !=| BIGINT-REQUIRED,BIGINT-REQUIRED | BIT |
built-in | false |
| < | BIGINT-REQUIRED,BIGINT-REQUIRED | BIT |
built-in | false |
| <=| BIGINT-REQUIRED,BIGINT-REQUIRED | BIT |
built-in | false |
| <>| BIGINT-REQUIRED,BIGINT-REQUIRED | BIT |
built-in | false |
| = | BIGINT-REQUIRED,BIGINT-REQUIRED | BIT |
built-in | false |
| ==| BIGINT-REQUIRED,BIGINT-REQUIRED | BIT |
built-in | false |
| > | BIGINT-REQUIRED,BIGINT-REQUIRED | BIT |
built-in | false |
| >=| BIGINT-REQUIRED,BIGINT-REQUIRED | BIT |
built-in | false |
| equal | BIGINT-REQUIRED,BIGINT-REQUIRED | BIT |
built-in | false |
| greater_than | BIGINT-REQUIRED,BIGINT-REQUIRED | BIT |
built-in | false |
| greater_than_or_equal_to | BIGINT-REQUIRED,BIGINT-REQUIRED | BIT |
built-in | false |
| less_than | BIGINT-REQUIRED,BIGINT-REQUIRED | BIT |
built-in | false |
| less_than_or_equal_to | BIGINT-REQUIRED,BIGINT-REQUIRED | BIT |
built-in | false |
| not_equal | BIGINT-REQUIRED,BIGINT-REQUIRED | BIT |
built-in | false |
+---+--+-+---+---+
14 rows selected (0.282 seconds)
{code}
So, I'm not sure this would qualify as a bug.
[~arina] do you agree?
> sys.functions table needs a fix in the names column
> ---
>
> Key: DRILL-6937
> URL: https://issues.apache.org/jira/browse/DRILL-6937
> Project: Apache Drill
> Issue Type: Bug
> Components: Execution - Flow
>Affects Versions: 1.15.0
>Reporter: Khurram Faraaz
>Assignee: Kunal Khatua
>Priority: Minor
> Fix For: 1.16.0
>
>
> The function names in the name column of sys.functions in some cases, are the
> operators, this is not the expected behavior, the name column should have
> actual names and not the operators.
> I am on Drill 1.15.0 commit : 8743e8f1e8d5bca4d67c94d07a8560ad356ff2b6
> {noformat}
> Apache Drill 1.15.0
> "Data is the new oil. Ready to Drill some?"
> 0: jdbc:drill:schema=dfs.tmp> select count(*) from sys.functions;
> +-+
> | EXPR$0 |
> +-+
> | 2846 |
> +-+
> 1 row selected (0.327 seconds)
> 0: jdbc:drill:schema=dfs.tmp>
> {noformat}
> {noformat}
> 0: jdbc:drill:schema=dfs.tmp> select distinct name from sys.functions limit
> 12;
> ++
> | name |
> ++
> | != |
> | $sum0 |
> | && |
> | - |
> | /int |
> | < |
> | <= |
> | <> |
> | = |
> | == |
> | > |
> | >= |
> ++
> 12 rows selected (0.175 seconds)
> {noformat}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Commented] (DRILL-6937) sys.functions table needs a fix in the names column
[
https://issues.apache.org/jira/browse/DRILL-6937?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16730499#comment-16730499
]
Kunal Khatua commented on DRILL-6937:
-
[~khfaraaz] Drill operators are also implicitly functions, which is why they
appear in the listing. That is how Drill is able to enforce data type
compatibility, because those operators get translated to functions with valid
arguments.
That said, some functions are internal in nature, e.g.
{code}
0: jdbc:drill:drillbit=10.10.100.127> select distinct name from sys.functions
where internal limit 3;
+---+
| name |
+---+
| castEmptyStringNullableVar16CharToNULLABLEBIGINT |
| castEmptyStringNullableVar16CharToNULLABLEDATE |
| castEmptyStringNullableVar16CharToNULLABLEFLOAT4 |
+---+
3 rows selected (0.408 seconds)
{code}
However, the {{$sum0}} and {{/int}} are operators I've not seen before and
might be part of some new UDFs. I'll take a look at it and update the JIRA with
the appropriate resolution.
> sys.functions table needs a fix in the names column
> ---
>
> Key: DRILL-6937
> URL: https://issues.apache.org/jira/browse/DRILL-6937
> Project: Apache Drill
> Issue Type: Bug
> Components: Execution - Flow
>Affects Versions: 1.15.0
>Reporter: Khurram Faraaz
>Assignee: Kunal Khatua
>Priority: Major
> Fix For: 1.16.0
>
>
> The function names in the name column of sys.functions in some cases, are the
> operators, this is not the expected behavior, the name column should have
> actual names and not the operators.
> I am on Drill 1.15.0 commit : 8743e8f1e8d5bca4d67c94d07a8560ad356ff2b6
> {noformat}
> Apache Drill 1.15.0
> "Data is the new oil. Ready to Drill some?"
> 0: jdbc:drill:schema=dfs.tmp> select count(*) from sys.functions;
> +-+
> | EXPR$0 |
> +-+
> | 2846 |
> +-+
> 1 row selected (0.327 seconds)
> 0: jdbc:drill:schema=dfs.tmp>
> {noformat}
> {noformat}
> 0: jdbc:drill:schema=dfs.tmp> select distinct name from sys.functions limit
> 12;
> ++
> | name |
> ++
> | != |
> | $sum0 |
> | && |
> | - |
> | /int |
> | < |
> | <= |
> | <> |
> | = |
> | == |
> | > |
> | >= |
> ++
> 12 rows selected (0.175 seconds)
> {noformat}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
