@StepScope should default to ScopedProxyMode.TARGET_CLASS [BATCH-2023]

[spring-projects-issues]

Tobias Flohre opened BATCH-2023 and commented

If I use @StepScope on a @Bean method that returns a FlatFileItemReader, I get a ClassCastException: java.lang.ClassCastException: $Proxy16 cannot be cast to org.springframework.batch.item.file.FlatFileItemReader

That's because the ScopedProxyMode is INTERFACES, and the proxy created is based on of the interfaces of FlatFileItemReader, and not FlatFileItemReader itself. Since we always have CGLIB when using @Configuration, I see no reason for not changing the default proxyMode of @StepScope to TARGET_CLASS.

One more argument for the proxyMode TARGET_CLASS: when I change the return type of the @Bean method to ItemReader<xx>, but still return a FlatFileItemReader, I don't get a ClassCastException, but the ItemStream registration does not work, and the FlatFileItemReader throws an exception on the first read.

Altogether those are serious stumble blocks for people not familiar with the proxying mechanisms in Spring, and we would avoid all of them with the default proxyMode TARGET_CLASS.

---

Affects: 2.2.0.RC1

Sub-tasks:

• BATCH-2822 Class cast exception when autowiring flat file writer

Issue Links:

• BATCH-1988 Refactor @StepScope to allow proxyMode = ScopedProxyMode.TARGET_CLASS
  ("duplicates")

Referenced from: commits 3780068

Backported to: 2.2.0.RC2

[spring-projects-issues]

Dave Syer commented

Initially I thought that backward compatibility with the old XML would be more important, but on reflection I think you are right: you always have cglib on the classpath now anyway (since Spring 3.2 carries it inside spring-core), so the default could be TARGET_CLASS and no-one would even notice.

[spring-projects-issues]

Dave Syer commented

Done. As predicted there was no sign of anything breaking.

[spring-projects-issues]

rovshenn commented

This bug is still occurring in the version for the following scenario. It works if I construct a job using Java with Job builder factory, but when I try to run the job using xml file it throws the similar exception as in the bug: This bug is still occurring in the version for the following scenario. It works if I construct a job using Java with Job builder factory, but when I try to run the job using xml file it throws the similar exception as in the bug:

ClassCastException: com.sun.proxy.$Proxy97 cannot be cast to 
org.springframework.batch.item.support.AbstractItemCountingItemStreamItemReader

<groupId>org.springframework</groupId>
<artifactId>spring-core</artifactId>
<version>5.1.4.RELEASE</version>

<groupId>org.springframework.boot</groupId>
<artifactId>spring-boot</artifactId>
<version>2.0.8.RELEASE</version>

<groupId>org.springframework.batch</groupId>
<artifactId>spring-batch-core</artifactId>
<version>4.1.1.RELEASE</version>

 

@Configuration
@PropertySource({"classpath:batch/common-${env}.properties", "classpath:batch/${job}-job.properties"})
@EnableBatchProcessing
public class SampleBatchConfig {

	public static final boolean IGNORE_INVALID_RECORDS = true;

	@Bean
	@StepScope
	public FlatFileItemWriter<Retirement> itemWriter(@Value("#{jobParameters['output.file.name']}") String filePath) {

		BeanWrapperFieldExtractor<Retirement> beanWrapperFieldExtractor = new BeanWrapperFieldExtractor<>();
		beanWrapperFieldExtractor.setNames(new String[]{"employeeId", "yearsToRetire"});
		String delimiter = "\t";
		return new FlatFileItemWriterBuilder<Retirement>()
				.resource(new FileSystemResource(filePath))
				.headerCallback(writer -> { writer.write("employeeId" + delimiter + "yearsToRetire"); })
				.name("itemWriter")
				.encoding(StandardCharsets.UTF_8.name())
				.shouldDeleteIfExists(true)
				.delimited()
				.delimiter(delimiter)
				.fieldExtractor(beanWrapperFieldExtractor)
				.build();
	}

}

Sample job in xml where we want to inject the flat file item writer bean

<?xml version="1.0" encoding="UTF-8"?>
<beans xmlns="http://www.springframework.org/schema/beans"
       xmlns:batch="http://www.springframework.org/schema/batch"
       xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
       xmlns:context="http://www.springframework.org/schema/context"
       xsi:schemaLocation="http://www.springframework.org/schema/batch http://www.springframework.org/schema/batch/spring-batch.xsd
       http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans-3.2.xsd 
       http://www.springframework.org/schema/context http://www.springframework.org/schema/context/spring-context.xsd">
    <context:component-scan base-package="com.test.batch"/>
    <batch:job id="job" restartable="true" incrementer="">
        <!--sample step that reads data from csv file processes values and writes new objects to a tsv-->
        <batch:step id="sampleStep2">
            <batch:tasklet>
                <batch:chunk reader="itemReader" processor="itemProcessor"
                             writer="itemWriter"
                             commit-interval="${step.chunkSize:1000}"/>
            </batch:tasklet>
        </batch:step>
    </batch:job>
</beans>