You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@pig.apache.org by "Koji Noguchi (Jira)" <ji...@apache.org> on 2020/10/12 19:35:00 UTC
[jira] [Commented] (PIG-5243) describe with typecast on as-clause
shows the types before the typecasting
[ https://issues.apache.org/jira/browse/PIG-5243?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17212625#comment-17212625 ]
Koji Noguchi commented on PIG-5243:
-----------------------------------
This bug made the debugging harder in PIG-5404. Looking.
This is happening because we have
{code:java|title=PigServer.Graph}
private Map<String, Operator> operators = new HashMap<String, Operator>();
...
private LogicalPlan lp;
...
{code}
{{operators}} and {{lp}} are both initialized at
{code:java}
1827 QueryParserDriver parserDriver = new QueryParserDriver( pigContext, scope, fileNameMap );
1828 lp = parserDriver.parse( query );
1829 operators = parserDriver.getOperators();
1830 lastRel = parserDriver.getLastRel();
{code}
These {{operators}} and {{lp}} are consistent at this point.
But later, we update the logical plan later by calling
{code:java}
1859 private void compile() throws IOException {
1860 lp.validate(pigContext, scope, false);
1861 ...
1862 }
{code}
As part of this LogicalPlan.validate(), we split the as-clause foreach into two separate foreach (PIG-2315).
(a) Original Foreach
followed by a
(b) TypeCast Foreach.
It is only this (b) TypeCast Foreach that is keeping the user's declared types but
PigServer.Graph.operators is still pointing to (a) Original Foreach.
> describe with typecast on as-clause shows the types before the typecasting
> --------------------------------------------------------------------------
>
> Key: PIG-5243
> URL: https://issues.apache.org/jira/browse/PIG-5243
> Project: Pig
> Issue Type: Bug
> Reporter: Koji Noguchi
> Assignee: Koji Noguchi
> Priority: Minor
>
> For code like
> {code}
> a = load 'test.txt' as (mytuple:tuple (), gpa:float);
> b = foreach a generate mytuple as (mytuple2:(name:int, age:double));
> store b into '/tmp/deleteme';
> {code}
> {{describe b}} shows
> {noformat}
> b: {mytuple2: (name: bytearray,age: bytearray)}
> {noformat}
> Execution wise, it is fine since there is an extra foreach typecasting the above relation.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)