Welcome!

Machine Learning Authors: Pat Romanski, Zakia Bouachraoui, Yeshim Deniz, Elizabeth White, Liz McMillan

Related Topics: Java IoT, Industrial IoT, Microservices Expo, IBM Cloud, Weblogic, Machine Learning

Java IoT: Article

Java Method Size

Getting around the 64K limit

The Java Virtual Machine specification limits the size of generated Java byte code for each method in a class to the maximum of 64K bytes. This limitation will cause the JVM to throw java.lang.VerifyError at runtime when the method size exceeds this limit.

This method size restriction of the JVM seems to be too stringent. A bug is open with the Oracle Java development team (http://bugs.sun.com/view_bug.do?bug_id=4262078). This bug is taken as an enhancement request which is likely to be fixed in the future java versions. That said, the fix for this issue is rather supportive. Application/Container developers need to consider few factors and apply good programming/refactoring techniques to get around this problem. While future versions of JVM might increase the method size limit to a considerable extent, a the size will always be a finite value.

In this article, I discuss some of the techniques that can be used to get around this problem. The techniques/suggestions discussed here, while apply to development environments also, are primarily targeted for containers which involve code generation as part of compilation process (e.g., EJB container or a JSP container).

Applications and Method Size Limit
While applications written in Java seldom encounter this issue, even more so after the above noted bug is fixed, it is recommended to follow the proven software engineering principles like code refactoring and object oriented programming techniques.

The complexity of this issue in application development is relatively smaller. As the code is written through the development phase, a number of tests are run both at the Unit level as well as the system level. Also, it is very rare that the application development involves complex code generation during the build time. This gives the developer/architect an opportunity to identify the possible method size restriction issue well ahead of the final delivery.

Application Containers and Method Size Limit
Application Containers are environments where Java applications execute. So they carry very high degree of complexity in terms of code. As the containers consume the application code and configuration and provide services for the application runtime, containers most often involve code generation. This code generation can be very complex in systems like JSP compiler as they convert scripting languages to Java.

This makes containers involving code generation particularly vulnerable to the Java method size limit restriction.

Techniques for Resolving the Method Size Restriction
In this section, I detail some of the techniques which can be used in code generation area to cover most of the use cases which cause the method size restriction.

Splitting a Big Method into Smaller Pieces
This is one of the structured programming techniques which can be used with effect. The driving factor for this technique should be the code refactoring. After the code generation is completed, a careful examination of the bigger methods can give us some clue of moving some of the java statements to a different private method. This is typically reverse engineering the output to a fine grained control.

Using a Intermediate Super Class
This applies particularly to areas such as EJB and JSP. A JSP file once converted into a Java class will typically extend a JspServlet class (implemented differently by different container vendors). The core refactoring technique here is to move the rudimentary code into the super class so that the generated class becomes light. However, this technique is fairly primitive. Most modern container vendors have already optimized their code to include almost everything possible into their super class implementation.

Using Helper Class/Methods
The third refactoring technique is to use a helper class. One particular use case is a JSP compiler handling a %@include...% directive. JSP specification mandates that the static includes need to be expanded inline in the generated service method. This adds to the size of the generated service method. One refactoring that can be used here is to move the expanded code into a private member function in the generated Java file and include just the method call in the service method.

Another area is the scripting elements in the JSP file. All the scripting elements are parsed and then included inline in the generated service() method of the java file. A similar refactoring technique discussed above can be used for the scripting elements as well.

Code generation engines can consider generating a Helper class which contains all the refactored methods and shield the Helper class from end-user's direct access.

Using JVM Switches
Java compilers generate .class files, with each file containing bytecodes for the methods in one class. These files contain a series of sections, or attributes, and it's possible to exercise some control over which sections actually show up in a .class file. The Code attribute contains actual bytecodes for methods, SourceFile information on the source file name used to generate this .class file, LineNumberTable mapping information between bytecode offsets and source file line numbers, and LocalVariableTable mapping information between stack frame offsets and local variable names.

We can use options to the javac compiler to control which sections are included:

javac Code, SourceFile, LineNumberTable
javac -g Code, SourceFile, LineNumberTable, LocalVariableTable
javac -g:none Code

How much do these options affect code size? A test using the JDK 1.2.2 sources in the java.io, java.lang, and java.util packages was run, with total sizes of all .class files as follows:

javac 668K
javac -g 815K
javac -g:none 550K

So use of javac -g:none saves about 20% in space over the default javac, and using javac -g is about 20% larger than the default. There are also commercial tools that tackle this particular problem.

If you strip out information such as line numbers from .class files, you may have problems with debugging and exception reporting. So it is recommended that the container vendors provide a configuration mechanism to switch off/on this feature so the production environments can take advantage.

Conclusion
Any refactoring technique will not completely alleviate the method size restriction. Excessive template text and scripting elements in the scripting languages will almost always cause the compilers to generate heavier methods and may lead to the method size restriction. The solution to this problem is supportive. Developers need to concentrate on good programming techniques so that the scripting language code like JSP will be more maintainable. JSP specification recommends using tag libraries that promote reusability while considerably reducing the size of the generated service method.

More Stories By Shankar Itchapurapu

Shankar Itchapurapu is a software engineer at Oracle in India. He holds a Master's degree in Computer Applications. You can e-mail Shankar at [email protected]

CloudEXPO Stories
The Transparent Cloud-computing Consortium (T-Cloud) is a neutral organization for researching new computing models and business opportunities in IoT era. In his session, Ikuo Nakagawa, Co-Founder and Board Member at Transparent Cloud Computing Consortium, will introduce the big change toward the "connected-economy" in the digital age. He'll introduce and describe some leading-edge business cases from his original points of view, and discuss models & strategies in the connected-economy. Nowadays, "digital innovation" is a big wave of business transformation based on digital technologies. IoT, Big Data, AI, FinTech and various leading-edge technologies are key components of such business drivers.
"DevOps is set to be one of the most profound disruptions to hit IT in decades," said Andi Mann. "It is a natural extension of cloud computing, and I have seen both firsthand and in independent research the fantastic results DevOps delivers. So I am excited to help the great team at @DevOpsSUMMIT and CloudEXPO tell the world how they can leverage this emerging disruptive trend."
Having been in the web hosting industry since 2002, dhosting has gained a great deal of experience while working on a wide range of projects. This experience has enabled the company to develop our amazing new product, which they are now excited to present! Among dHosting's greatest achievements, they can include the development of their own hosting panel, the building of their fully redundant server system, and the creation of dhHosting's unique product, Dynamic Edge.
Nutanix has been named "Platinum Sponsor" of CloudEXPO | DevOpsSUMMIT | DXWorldEXPO New York, which will take place November 12-13, 2018 in New York City. Nutanix makes infrastructure invisible, elevating IT to focus on the applications and services that power their business. The Nutanix Enterprise Cloud Platform blends web-scale engineering and consumer-grade design to natively converge server, storage, virtualization and networking into a resilient, software-defined solution with rich machine intelligence.
Concerns about security, downtime and latency, budgets, and general unfamiliarity with cloud technologies continue to create hesitation for many organizations that truly need to be developing a cloud strategy. Hybrid cloud solutions are helping to elevate those concerns by enabling the combination or orchestration of two or more platforms, including on-premise infrastructure, private clouds and/or third-party, public cloud services. This gives organizations more comfort to begin their digital transformation without a complete overhaul of their existing infrastructure - serving as a sort of "missing link" for transition to cloud utilization.