Tuesday, April 18, 2017

Oracle JDK 9 Early Access Documentation Updated

Raymond Gallardo's 4 April 2017 post Early Access documentation for Oracle JDK 9 has been updated today announces updates to the Oracle JDK9 Documentation Early Access page. Gallardo highlights a few of the updated sections including What's New in Oracle JDK 9, Oracle JDK 9 Migration Guide, HotSpot Virtual Machine Garbage Collection Tuning Guide (including Garbage-First Garbage Collector Tuning), javapackager tool for "packaging Java and JavaFX applications," and XML Catalog API.

The Main Tools to Create and Build Applications section features information on jlink (JEP 282), jmod (create and show contents of Project Jigsaw JMOD files), and jdeprscan (static analysis tool that scans ... for uses of deprecated API elements). There is also information on jhsdb for obtaining "specific information from a hanging or crashed JVM."

I summarized some of the tools and features being removed from the JDK with Java 9 in the post JDK 9 is the End of the Road for Some Features. Some of these are spelled out with additional details in the section Removed Tools. The Changes to Garbage Collection section similarly shows changes and removals in Java 9 of options and commands related to garbage collection. The section Changes to the Installed JDK/JRE Image outlines removal of things such as rt.jar and tools.jar, the extension mechanism, and the endorsed standards override mechanism.

The section Enable Logging with the JVM Unified Logging Framework demonstrates how to "use -Xlog option to configure or enable logging with the Java Virtual Machine (JVM) unified logging framework." The Changes to GC Log Output section explains how to use this -Xlog with gc (-Xlog:gc) for logging related to garbage collection and points out that "the -XX:+PrintGCDetails and -XX:+PrintGC options have been deprecated."

I previously blogged on the future of the Concurrent-Mark-Sweep garbage collector. The Java SE 9 HotSpot Virtual Machine Garbage Collection Tuning Guide states, "The CMS collector is deprecated. Strongly consider using the Garbage-First collector instead."

The main Oracle JDK 9 Documentation Early Access page is at http://docs.oracle.com/javase/9/index.html and its Developer Guides link provides direct references to guides such as Troubleshooting Guide, HotSpot Virtual Machine Garbage Collection Tuning Guide, Java Scripting Programming Guide, Java Language Updates, and Migration Guide.

Monday, April 17, 2017

Implications of the Presence of StringBuffer

When I am working on legacy code and run across instances of StringBuffer, I typically replace them with instances of StringBuilder. Although a performance advantage can be gained from this change, I often change it in places I know will have little noticeable effect in terms of performance. I feel it's worth making the change for a variety of reasons in addition to the potential for performance benefit. There's rarely a reason to not choose StringBuilder over StringBuffer (API expectations are the most common exception) and the existence of StringBuffer in code misleads and provides a bad example to those new to Java.

In the book The Pragmatic Programmer: From Journeyman to Master, Andy Hunt and David Thomas discuss "the importance of fixing the small problems in your code, the 'broken windows'." Jeff Atwood touched on this subject in the post The Broken Window Theory and it has been more recently addressed in the posts Software Rot, Entropy and the Broken Window Theory and Don't leave broken windows. The presence of StringBuffer implies a staleness in the code. In effect, use of StringBuffer may not be a "broken window," but it's a really old, leaky single-pane window that should be replaced with a modern, energy-efficient double-pane window.

I found Peter Lawrey's recent blog post StringBuffer, and how hard it is to get rid of legacy code to be an interesting take on other implications of StringBuffer that still exist in code. Lawrey quotes the last paragraph of the StringBuffer class Javadoc documentation, "As of release JDK 5, this class has been supplemented with an equivalent class designed for use by a single thread, StringBuilder. The StringBuilder class should generally be used in preference to this one, as it supports all of the same operations but it is faster, as it performs no synchronization." Lawrey then uses simple Java methods and jmap to demonstrate that instances of StringBuffer are still used in classes and libraries delivered with the JDK even as late as Java 8.

Lawrey points out that the presence of StringBuffer in frequently used Java code more than a decade after the introduction of "drop-in replacement" StringBuilder is evidence of how difficult it is to "clean up legacy code." Lawrey's full conclusion states, "Using StringBuffer on startup doesn’t make much difference, but given it has a well known, drop in replacement, and it is still used, even in new functionality more than ten years later shows how hard it can be to clean up legacy code or to change thinking to get people to use best practice libraries."

I decided to try out one of Lawrey's simplest examples when compiled with Java 8 Update 121 and when compiled with a recent release of OpenJDK 9. I (slightly) adapted Lawrey's example to the simple "Main" class listing shown next.


import java.io.IOException;

 * (Slightly) adapted class from blog post
 * "StringBuffer, and how hard it is to get rid of legacy code" at
 * https://vanilla-java.github.io/2017/04/13/String-Buffer-and-how-hard-it-is-to-get-rid-of-legacy-code.html
public class Main
    * Main function that instantiates this Java "application" and does nothing
    * else until "ENTER" is pressed.
   public static void main(final String[] args) throws IOException
      System.out.println("Waiting [press ENTER to exit] ..");

The following screen snapshot shows the output of using jcmd with its -all option (includes unreachable objects in the inspection) to show the number of instances of StringBuffer and StringBuilder in the simple Java application when compiled and run against three different versions of Java (Java 8 Update 102, Java 8 Update 121, and OpenJDK 9.0 ea+164). The execution of jcmd is performed in PowerShell and so Select-String is used similarly to how grep is used in Linux.

Although the versions of the class compiled and executed with versions of Java 8 had instances of StringBuffer, the version compiled with and executed against Java 9 only had instances of StringBuilder. It looks like the resolution of JDK-8041679 ("Replace uses of StringBuffer with StringBuilder within core library classes") and JDK-8043342 ("Replace uses of StringBuffer with StringBuilder within crypto code") have had their intended effect.

Wednesday, April 12, 2017

Use Cases for Java Enhanced Enums

In the message Enhanced Enums -- use cases, Brian Goetz writes, "We're hoping to get user feedback on the feature [Enhanced Enums] as it is now implemented." He states the first purpose of his message, "To get things started, here are some typical use cases where generic enums might be useful." The first of the two presented examples is refactoring com.sun.tools.javac.code.Dynamic class and its eight factory methods returning different instances of BootstrapArgument with different instances of its nested Kind enum into a single method using the dynamic enum.

The second use case example of a possible application of enhanced enums that Goetz provides is command line parsing in which an enum is used to represent the data types of parameters. Vicente Romero replied to Goetz's message with two more examples of where enhanced enums might be applied: "code sharing between enum constants" and "the power of sharper typing".

Goetz encourages others to provide more use cases for enhanced enums, "Please contribute others, as well as places in the JDK where code could be refactored using this feature." He concludes, "If anyone wants to experiment and offer their experience in applying (or misapplying) this feature, either to the JDK or their own codebase, that would be appreciated...."

Tuesday, April 11, 2017

Java Finalizer and Java File Input/Output Streams

I often find myself noticing topics online more after I've worked directly with them or spent time learning about them. The recent Stephen Connolly (CloudBees) post FileInputStream / FileOutputStream Considered Harmful caught my attention because of my recent issues with Java's finalizer. In that post, the author talks about potential consequences of java.io.FileInputStream and java.io.FileOutputStream implementing overridden finalize() methods FileInputStream.finalize() and FileOutputStream.finalize(). With talk of deprecating the finalizer in JDK 9, my perspective is that a subject I had not thought about in years is all of a sudden all around me.

Connolly's post references the Hadoop JIRA HDFS-8562 ("HDFS Performance is impacted by FileInputStream Finalizer"). That JIRA was opened in June 2015 and its description includes interesting background on why the finalizer of FileInputStream causes issues for those using HDFS. This JIRA is also interesting because it looks at why it's not trivial to change FileInputStream and FileOutputStream to not use the protected finalize() methods.

JDK-8080225 ("FileInputStream cleanup should be improved.") is referenced in HDFS-8562 and was written in May 2015. It states, "FileInputStream relies on finalization to perform final closes if the FIS is not already closed. This results in extra work for GC that occurs in a burst. The cleanup of FileInputStreams should happen sooner and not contribute to overhead in GC." Alan Bateman has commented on this with a work-around, "The issue can be easily worked around by using Files.newInputStream instead." Roger Riggs writes of the difficulty of adequately addressing this issue, "Since it is unknown/unknowable how many FIS/FOS subclasses might rely on overriding close or finalize the compatibility issue is severe. Only a long term (multiple release) restriction to deprecate or invalidate overriding would have possibility of eventually eliminating the compatibility problem."

Connolly ends his post with reference to Jenkins changing this via JENKINS-42934 ("Avoid using new FileInputStream / new FileOutputStream"). An example of changing new FileInputStream to Files.newInputStream is available from there.

The fact that I've been able to use Java for so many years without worrying about the finalizer even while I used classes such as FileInputStream is evidence that, by themselves, limited use of these classes with finalize() implementations doesn't necessarily lead to garbage collection or other problems. I like how Colin P. McCabe articulates the issue in the HDFS JIRA on this: "While it's true that we use FileInputStream / FileOutputStream in many places, most of those places have short-lived objects or only use very small numbers of objects. Like I mentioned earlier, the big problem with finalizers we encounter is in the short-circuit read stream cache. If we can fix that, as this patch attempts to do, we will have eliminated most of the problem." In other words, not all uses of FileInputStream and FileOutputStream are causes for concern. Using tools to identify unusually high garbage collection related to finalizers is the best way to identify those that need to be addressed.

For many years of Java development, I did not use or case about the Java finalizer. In recent months, it has become an issue that I'm seeing more people dealing with. Deprecating the Java finalizer is a good first step toward removing it from core APIs.

Monday, April 10, 2017

Java Garbage Collectors: When Will G1GC Force CMS Out?

In JEPs proposed to target JDK 9 (2017/4/4), Mark Reinhold has written that JEP 291 ("Deprecate the Concurrent Mark Sweep (CMS) Garbage Collector") is one of two JEPs that "have been placed into the 'Proposed to Target' state by their owners after discussion and review". If things go well for JEP 291, it will be targeted for JDK 9.

Reinhold explains in this message why JEP 291 can still be targeted to JDK 9 at this relatively late date: "JEP 291 requires only a minuscule code change, to enable the proposed warning message to be issued. It's a JEP in the first place not because it's a risky change but to bring visibility to the plan to remove the CMS collector in the long term." As these sentences state, the JDK 9 targeted action is simply to mark the Concurrent Mark Sweep (CMS) collector as deprecated with the idea that it will be removed at some point "in the long term."

Although G1GC is the default garbage collector for JDK 9 through JEP 248, it is not always the best collector for all situations. Even the proposal to deprecate CMS acknowledges this in its "Risks and Assumptions" which states, "For some applications CMS is a very good fit and might always outperform G1."

Another recent discussion of OpenJDK jdk9-dev mailing list is titled "JEP 291: Deprecate the Concurrent Mark Sweep (CMS) Garbage Collector" contains interesting arguments for retaining CMS. Christoph Engelbert (Hazelcast) writes, "CMS+ParNew is the most commonly deployed solution and a lot of applications are optimized to the behavior of CMS." Scott Palmer writes that in his specific application, "we have found that so far the CMS collector has significantly lower maximum pause times than G1." Roman Kennke (RedHat) adds, "I'd say it's too early to talk about removing CMS. And, to be honest, I even question the move to deprecate it." Martijn Verburg (jClarity) states, "We are now constantly asked to tune G1 for customers and have found that even with our most advanced analytics (in combination with some of the common and more esoteric tuning options), we are unable to get G1 to outperform CMS for *certain* use cases. Several customers have therefore reverted to CMS and are very interested in its future (as consumers)."

This same discussion also includes reasons for deprecating CMS. Mark Reinhold's post states that JEP 291 was "posted last summer" and requests were made for a CMS maintainer, but "so far, no one has stepped up." He concludes that post, "In any case, Oracle does intend to stop maintaining CMS at some point in the not-to-distant future, and if no one ever steps up then we'll remove the code."

Jeremy Manson (Google) explains the trickiness of the current situation with G1GC and CMS:

We decided that supporting CMS in any sort of ongoing fashion should be a last resort after we try getting G1 to do what we need it to do. Our belief is that fewer collectors is better. We spent some time over the last few months coordinating with some of the folks at Oracle and experimenting to see if there were plausible ways forward with G1. We couldn't find anything obvious.

The gist of all this seems to be that many applications still depend on CMS and these applications will have a deprecation warning displayed in JDK 9. The future of the CMS garbage collector appears to be in doubt, but it would only be deprecated in JDK 9. When the CMS collector would actually be removed seems less obvious, but I assume that JDK 10 is a potential "future major release" in which CMS support could be terminated. Quoting Manson (Google) again, "The short of it is: We are still willing to contribute work to support CMS, but we want to make sure we've done our due diligence on G1 first. Our belief has been that the JDK 10 timeframe is long enough that we don't have to make this decision hurriedly."

It looks likely that Java applications using the Concurrent Mark Sweep garbage collector in JDK9 will see warning messages about the deprecation of the CMS garbage collector. When (or if) CMS won't be available at all is less obvious and depends on who is willing to continue supporting CMS.

Saturday, April 1, 2017

How to Effectively Sweep Problems Under the Rug in Java

Because software bugs can make us appear bad as developers and lead to others thinking less of us, it is best to avoid writing bugs, to identify and fix bugs quickly, or to cover up our bugs. There are numerous blog posts and articles that discuss avoiding bugs and identifying and fixing bugs, so, in this blog post, I look at some of the most effective tactics for sweeping problems in Java code bases under the rug.

Swallow Checked Exceptions

Exceptions are one of the things that give us away when we've accidentally introduced bugs into the code. It's also a hassle to declare a throws clause on a method or catch a checked Exception. A solution to both problems is to simply catch the exception (even if it's a RuntimeException) at the point it might be thrown and do nothing. That keeps the API concise and there is little one could do about the checked exception anyway. By not logging it or doing anything about it, no one even needs to know it ever happened.

Comment Out or Remove Unhappy Unit Tests

Failed unit tests can be distracting and make it difficult to determine when new functionality has broken the tests. They can also reveal when we've broken something with code changes. Commenting out these failing unit tests will make the unit test report cleaner and make it easier to see if anyone else's new functionality breaks the unit tests.

Use @Ignore on JUnit-based Unit Tests

It may seem distasteful to comment out failing unit tests, so another and possibly more pleasing alternative is to annotate failing JUnit-based unit test methods with the @Ignore annotation.

Remove Individual Tests Altogether

If commenting out a broken test or annotating a broken test with @Ignore are unsatisfactory because someone might still detect we've broken a unit test, we can simply remove the offending unit test altogether.

Comment Out the Offending Assertion

We don't necessarily need to comment out or remove entire tests. It can be as simple as commenting out or removing the assert statements in a unit test method. The method can execute and run successfully every time because no assertions means no way to fail. This is especially effective when the unit test methods are very long and convoluted so that lack of assertions is not easily spotted.

Distract with the Noise of Useless and Redundant Tests

Commenting out unit tests, annotating JUnit-based unit tests with @Ignore, and even removing unit tests might be too obvious of ploys for sweeping issues under the rug in Java. To make these less obvious, another effective tactic is to write numerous unnecessary and redundant test methods in the same unit test class so that it appears that comprehensive testing is being done, but in reality only a small subset of functionality (the subset we know is working) is being tested.

Write Unit Tests That 'Prove' Your Code is Correct Even When It Is Not

We can take advantage of the fact that unit tests can only test what the author of the unit test thinks is the expected behavior of the software under test to write unit tests that demonstrate our code to be correct. If our code for adding two integers together accidentally returns a sum of 5 when 2 and 2 are provided, we can simply set the expected result in the unit test to also be 5. A pretty unit test report is presented and no has to be the wiser.

Avoid Logging Details

Logs can expose one's software problems and an effective approach to dealing with this risk is to not log at all, only log routine operations and results, and leave details (especially context) out of logged messages. Excessive logging of mundane details can also obscure any more meaningful messages that might reveal our code's weaknesses.

Avoid Descriptive toString() Implementations

A descriptive toString() method can reveal far too much about the state of any given instance and reveal our mistakes. Not overriding Object.toString() can make it more difficult to identify issues and associate issues with any given code or developer. The extra time required to track down issues gives you more time to move onto the next project before it is discovered that it's your code that's at fault. If you write a Java class that extends a class with a descriptive toString() you can override that method in your extended class to do nothing (effectively removing the potentially incriminating toString() output). If you want it to appear as if it was never implemented at all in the inheritance hierarchy, be sure to have your extended class's toString() method return System.identityHashCode(this).

Don't Let NullPointerExceptions Betray You

The NullPointerException is probably the most common exception a Java developer deals with. These are especially dangerous because they often reveal code's weak spots. One tactic to to simply wrap every line of code with a try-catch and simply swallow the exception (including the NPE). Another and less obvious tactic is to avoid NPEs by never returning or passing a null. Sometimes there are obvious defaults that make sense to use instead of null (such as empty Strings or collections), but sometimes we have to be more creative to avoid null. This is where it can be useful to use a "default" non-null value in place of null. There are two schools of thought on how to approach this arbitrary non-null default. One approach is to use the most commonly seen value in the set of data as the default because, if it's common anyway, it may not be noticed when a few more of that value show up and you are more likely to have code that appears to process that common value without incident. On the other hand, if you have a value that is almost never used, that can make a good default because there may be less code (especially well tested code) affected by it than by the commonly expected value.


As I reviewed these tactics to sweep issues in Java code under the rug, I noticed some recurring themes. Exceptions, logging, and unit tests are particularly troublesome in terms of exposing our software's weaknesses and therefore it's not surprising that most of the ways of effectively "covering our tracks" relate to handling of exceptions, logging, and unit tests.




Happy April Fools' Day!