[
https://issues.apache.org/jira/browse/ASTERIXDB-2918?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17395295#comment-17395295
]
Wail Y. Alkowaileet commented on ASTERIXDB-2918:
------------------------------------------------
HI Ingo,
I have a problem with my terminology here. Sometimes we refer to declared
fields as closed fields :)
So, what I meant is no fields should be declared (no closed fields). When you
want to read from Parquet, the type should have no declared field as in the
undeclaredDataset. We should validate that during the creation of the dataset.
Sorry about the confusion!
> IndexOutOfBoundsException when querying Parquet files
> -----------------------------------------------------
>
> Key: ASTERIXDB-2918
> URL: https://issues.apache.org/jira/browse/ASTERIXDB-2918
> Project: Apache AsterixDB
> Issue Type: Bug
> Components: EXT - External data
> Reporter: Ingo Müller
> Assignee: Wail Y. Alkowaileet
> Priority: Major
> Attachments: Run2012B_SingleMu_restructured_1000.parquet,
> create_event_type.sqlpp, stacktrace.log
>
>
> I am getting an IndexOutOfBoundsException when creating an external table
> based on Parquet files onHDFS or loading them into an existing table if I
> specify a closed type for the table. If a specify an empty open type as
> follows, all works fine:
> {{CREATE TYPE anyType IF NOT EXISTS AS OPEN {};}}
> Then I create an external table as follows:
> CREATE EXTERNAL DATASET untypedDataset(anyType)
> USING hdfs
> (("hdfs"="hdfs://namenode:8020"),
> ("path"="/test/*.parquet"),
> ("input-format"="parquet-input-format"))
> With {{anyType}}, I can query the table just fine. However, if I use the
> {{eventType}} created as shown in the attachment, running any query against
> the dataset produces an error about an exception. In cc.log, I find the
> output as attached in {{stacktrace.log}}.
> I do not know how to debug this further.
> For your reference, I am using a self-compiled development from master from a
> few days ago (rev. 5120106e) running on AdoptOpenJDK 15. I am also attaching
> the Parquet file that caused the problem.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
