[
https://issues.apache.org/jira/browse/FLINK-30093?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17640143#comment-17640143
]
hubert dulay commented on FLINK-30093:
--------------------------------------
I may have found the solution to this. Basically the generated code doesn't
compile and the error message originating from codehaus.janino is a terrible
one. The line below is the culprit
[https://github.com/apache/flink/blob/master/flink-formats/flink-protobuf/src/main/java/org/apache/flink/formats/protobuf/util/PbFormatUtils.java#L39]
Changing it from:
{code:java}
return outerProtoName + descriptor.getName();{code}
{{to}}
{code:java}
return descriptor.getFile().getOptions().getJavaPackage() + "." +
descriptor.getName();{code}
Seemed to do the trick but I'm still working through it.
> [Flink SQL][Protobuf] CompileException when querying Kafka topic using
> google.protobuf.Timestamp
> -------------------------------------------------------------------------------------------------
>
> Key: FLINK-30093
> URL: https://issues.apache.org/jira/browse/FLINK-30093
> Project: Flink
> Issue Type: Bug
> Components: Formats (JSON, Avro, Parquet, ORC, SequenceFile), Table
> SQL / Ecosystem
> Affects Versions: 1.16.0
> Environment: Mac OS Ventura
> Reporter: James Mcguire
> Priority: Major
> Attachments: taskmanager_172.22.0 (1).4_46291-40eec2_log
>
>
> I am encountering an issue when trying to use Flink SQL to query a Kafka
> topic that uses {{{}google.protobuf.Timestamp{}}}.
>
> When attempting to use Flink SQL to query a protobuf serialized Kafka topic
> that uses {{{}google.protobuf.Timestamp{}}}, a
> {{org.codehaus.commons.compiler.CompileException: Line 23, Column 5: Cannot
> determine simple type name "com" }}error occurs when trying to query the
> table.
>
> *Replication steps:*
> 1. Use a protobuf definition that contains a
> {{{}google.protobuf.Timestamp{}}}:
> {noformat}
> syntax = "proto3";
> package example.message;
> import "google/protobuf/timestamp.proto";
> option java_package = "com.example.message";
> option java_multiple_files = true;
> message Test {
> int64 id = 1;
> google.protobuf.Timestamp created_at = 5;
> }{noformat}
> 2. Use protobuf definition to produce message to topic
> 3. Confirm message is deserializable by protoc:
> {code:java}
> kcat -C -t development.example.message -b localhost:9092 -o -1 -e -q -D "" |
> protoc --decode=example.message.Test
> --proto_path=/Users/jamesmcguire/repos/flink-proto-example/schemas/
> example/message/test.proto
> id: 123
> created_at {
> seconds: 456
> nanos: 789
> }{code}
> 4. Create table in Flink SQL using kafka connector and protobuf format
> {code:java}
> CREATE TABLE tests (
> id BIGINT,
> created_at row<seconds BIGINT, nanos INT>
> )
> COMMENT ''
> WITH (
> 'connector' = 'kafka',
> 'format' = 'protobuf',
> 'protobuf.message-class-name' = 'com.example.message.Test',
> 'properties.auto.offset.reset' = 'earliest',
> 'properties.bootstrap.servers' = 'host.docker.internal:9092',
> 'properties.group.id' = 'test-1',
> 'topic' = 'development.example.message'
> );{code}
> 5. Run query in Flink SQL and encounter error:
> {code:java}
> Flink SQL> select * from tests;
> [ERROR] Could not execute SQL statement. Reason:
> org.codehaus.commons.compiler.CompileException: Line 23, Column 5: Cannot
> determine simple type name "com" {code}
> {*}NOTE{*}: If you repeat steps 4-5 without {{created_at row<seconds BIGINT,
> nanos INT>}} in the table, step 5 will complete successfully.
> 6. Observe in attached log file, Flink appears to be using the incorrect
> namespace (should be {{google.protobuf.Timestamp):}}
> {code:java}
> com.example.message.Timestamp message3 = message0.getCreatedAt(); {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
