Frequently Asked Questions about SLF4J

Generalities

  1. What is SLF4J?
  2. When should SLF4J be used?
  3. Is SLF4J yet another logging facade?
  4. If SLF4J fixes Jakarta Commons Logging (JCL), then why wasn't the fix made in JCL instead of creating a new project?
  5. When using SLF4J, do I have to recompile my application to switch to a different logging system?
  6. What are SLF4J's requirements?
  7. What has changed in SLF4J version 2.0.0?
  8. Are SLF4J versions backward compatible?
  9. Is there a way to control the internal messages emitted by SLF4J?
  10. I am getting IllegalAccessError exceptions when using SLF4J. Why is that?
  11. When using the fluent API, how can I make my IDE alert me when I forget to call the log() method?
  12. Why is SLF4J licensed under X11 type license instead of the Apache Software License?
  13. Where can I get a particular SLF4J provider/binding?
  14. Should my library attempt to configure logging?
  15. In order to reduce the number of dependencies of our software we would like to make SLF4J an optional dependency. Is that a good idea?
  16. What about Maven transitive dependencies?
  17. How do I exclude commons-logging as a Maven dependency?
  18. What are the JPMS module names of the various SLF4J artifacts?
  19. What is the difference between an SLF4J binding and an SLF4J provider?
About the SLF4J API
  1. Why don't the printing methods in the Logger interface accept message of type Object, but only messages of type String?
  2. Can I log an exception without an accompanying message?
  3. What is the fastest way of (not) logging?
  4. How can I log the string contents of a single (possibly complex) object?
  5. Why doesn't the org.slf4j.Logger interface have methods for the FATAL level?
  6. Why was the TRACE level introduced only in SLF4J version 1.4.0?
  7. What are markers and what are they goood for?
  8. Does the SLF4J logging API support I18N (internationalization)?
  9. Is it possible to retrieve loggers without going through the static methods in LoggerFactory?
  10. In the presence of an exception/throwable, is it possible to parameterize a logging statement?
Implementing the SLF4J API
  1. How do I make my logging framework SLF4J compatible?
  2. How can my logging system add support for the Marker interface?
  3. How does SLF4J's version check mechanism work?
General questions about logging
  1. Should Logger members of a class be declared as static?
  2. Is there a recommended idiom for declaring a loggers in a class?

Generalities

What is SLF4J?

SLF4J is a simple facade for logging systems allowing the end-user to plug in the desired logging system at deployment time.

When should SLF4J be used?

In short, libraries and other embedded components should consider SLF4J for their logging needs because libraries cannot afford to impose their choice of logging framework on the end-user. On the other hand, it does not necessarily make sense for stand-alone applications to use SLF4J. Stand-alone applications can invoke the logging framework of their choice directly. In the case of logback, the question is moot because logback exposes its logger API via SLF4J.

SLF4J is only a facade, meaning that it does not provide a complete logging solution. Operations such as configuring appenders or setting logging levels cannot be performed with SLF4J. Thus, at some point in time, any non-trivial application will need to directly invoke the underlying logging system. In other words, complete independence from the API underlying logging system is not possible for a stand-alone application. Nevertheless, SLF4J reduces the impact of this dependence to near-painless levels.

Suppose that your CRM application uses log4j for its logging. However, one of your important clients request that logging be performed through java.util.Logging, a.k.a. JDK 1.4 logging. If your application is riddled with thousands of direct log4j calls, migration to java.util.logging would be a relatively lengthy and error-prone process. Even worse, you would potentially need to maintain two versions of your CRM software. Had you been invoking SLF4J API instead of log4j, the migration could be completed in a matter of minutes by replacing one jar file with another.

SLF4J lets component developers to defer the choice of the logging system to the end-user but eventually a choice needs to be made.

Is SLF4J yet another logging facade?

SLF4J is conceptually very similar to Apache Commons Logging a.k.a. Jakarta Commons Logging (JCL). As such, it can be thought of as yet another logging facade. However, SLF4J is much simpler in design and arguably more robust. In a nutshell, SLF4J avoid the class loader issues that plague JCL.

If SLF4J fixes Jakarta Commons Logging (JCL), then why wasn't the fix made in JCL instead of creating a new project?

This is a very good question. First, SLF4J static binding approach (prior to version 2.0) is very simple, perhaps even laughably so. It was not easy to convince developers of the validity of that approach. Note that as of version 2.0.0, SLF4J uses the ServiceLoader mechanism offered by the Java platform. This new approach is still relatively static and therefore predictable.

Second, SLF4J offers two enhancements which tend to be underestimated. Parameterized log messages solve an important problem associated with logging performance, in a pragmatic way. Marker objects, which are supported by the org.slf4j.Logger interface, pave the way for adoption of advanced logging systems and still leave the door open to switching back to more traditional logging systems if need be.

These days, the question is moot since JCL has been defunct for at least a decade.

When using SLF4J, do I have to recompile my application to switch to a different logging system?

No, you do not need to recompile your application. You can switch to a different logging system by removing the previous SLF4J provider and replacing it with the provider of your choice.

For example, if you were using the NOP implementation and would like to switch to reload4j, simply replace slf4j-nop.jar with slf4j-reload4j.jar on your classpath but do not forget to add reload4j.jar as well. Want to switch to java.util.logging? Just replace slf4j-reload4j.jar with slf4j-jdk14.jar.

What are SLF4J's requirements?

As of version 2.0.0, SLF4J requires JDK 8 or later. Version 1.7.0 requires JDK 1.5 or later.

 

Provider Requirements
slf4j-nop JDK 8
slf4j-simple JDK 8
slf4j-log4j (replaced by slf4j-reload4j) JDK 8, plus any other library dependencies required by the log4j/reload4j appenders in use
slf4j-jdk14 JDK 8 or later
logback-classic JDK 8 or later, plus any other library dependencies required by the logback appenders in use

What has changed in SLF4J version 2.0.0?

SLF4J 2.0.0 incorporates an optional fluent api. Otherwise, there are no client facing API changes in 2.0.x. For most users, upgrading to version 2.0..x should be a drop-in replacement, as long as the logging provider is updated as well.

In version 2.0.0, SLF4J has been modularized per JPMS/Jigsaw specification. The JPMS module names are listed in another FAQ entry.

More visibly, slf4j-api now relies on the ServiceLoader mechanism to find its logging backend. SLF4J 1.7.x and earlier versions relied on the static binder mechanism which is no longer honored by slf4j-api version 2.0.x. More specifically, when initializing the LoggerFactory class will no longer search for the StaticLoggerBinder class on the class path.

Instead of "bindings" now org.slf4j.LoggerFactory searches for "providers". These ship for example with slf4j-nop-2.0.x.jar, slf4j-simple-2.0.x.jar or slf4j-jdk14-2.0.x.jar.

The following table describes the results when various slf4j-api and slf4j-simple versions are placed on the class path. Please note that the table below applies by analogy not just to slf4j-simple but also to other providers such as slf4j-reload4j, logback-classic, slf4j-jdk14, etc...

slf4j-api version slf4j-simple version Result Explanation
2.0.x 2.0.x OK Same version for slf4j-api and provider
1.7.x 2.0.x no bindings can be found warning message 2.0.x providers do not act as 1.7.x/1.6.x compatible bindings
2.0.x 1.7.x no providers can be found warning message slf4j-api 2.0.x will no longer search for StaticLoggerBinding

since 2.0.9 You can specify the provider class explicitly via the "slf4j.provider" system property. This bypasses the service loader mechanism for finding providers and may shorten SLF4J initialization.

Are SLF4J versions backward compatible?

From the clients perspective, the SLF4J API, more specifically the org.slf4j package, is backward compatible for all versions. This means than you can upgrade from SLF4J version 1.0 to any later version without problems. Code compiled with slf4j-api-versionN.jar will work with slf4j-api-versionM.jar for any versionN and any versionM. To date, binary compatibility in slf4j-api has never been broken.

However, while the SLF4J API is very stable from the client's perspective, SLF4J providers, e.g. slf4j-simple.jar or slf4j-reload4j.jar, may require a specific version of slf4j-api. Mixing different versions of slf4j artifacts can be problematic and is strongly discouraged. For instance, if you are using slf4j-api-2.0.0.jar, then you should also use slf4j-simple-2.0.0.jar, using slf4j-simple-1.7.32.jar will not work.

At initialization time, if SLF4J suspects that there may be a version mismatch problem, it emits a warning about the said mismatch.

Fluent API requires version 2.0 If your code accesses the fluent API introduced in slf4j 2.0, then your code will require slf4j-api version 2.0 or later.

Is there a way to control the internal messages emitted by SLF4J?

As of version 2.0.10, SLF4J uses the Reporter class to output its internal messages. The Reporter mechanism is totally unrelated to the logging provider in use. Thus, it cannot be controlled by the SLF4J provider's settings, i.e. logback/log4j/jul settings have no effect on Reporter.

By default, Reporter sends its output to Stderr. However, by "slf4j.internal.report.stream" property to "System.out" "stdout" or "sysout", output can be redirected to stdout. Any other value will direct output to the default, i.e. stderr.

It is also possible to set the "slf4j.internal.verbosity" system property to one of "INFO", "WARN" or "ERROR" to control the verbosity of SLF4J internal messages. Mesages of level INFO are prefixed with the string "SLF4J(I). Similartly, level WARN is prefixed by SLF4J(W) and ERROR by SLF4J(E). ERROR messages cannot be supressed.

I am getting IllegalAccessError exceptions when using SLF4J. Why is that?

Here are the exception details.

Exception in thread "main" java.lang.IllegalAccessError: tried to access field
org.slf4j.impl.StaticLoggerBinder.SINGLETON from class org.slf4j.LoggerFactory
   at org.slf4j.LoggerFactory.<clinit>(LoggerFactory.java:60)

This error is caused by the static initializer of the LoggerFactory class attempting to directly access the SINGLETON field of org.slf4j.impl.StaticLoggerBinder. While this was allowed in SLF4J 1.5.5 and earlier, in 1.5.6 and later the SINGLETON field has been marked as private access.

If you get the exception shown above, then you are using an older version of slf4j-api, e.g. 1.4.3, with a new version of a slf4j binding, e.g. 1.5.6. Typically, this occurs when your Maven pom.ml file incorporates hibernate 3.3.0 which declares a dependency on slf4j-api version 1.4.2. If your pom.xml declares a dependency on an slf4j binding, say slf4j-log4j12 version 1.5.6, then you will get illegal access errors.

To see which version of slf4j-api is pulled in by Maven, use the maven dependency plugin as follows.

mvn dependency:tree

In your pom.xml file, explicitly declaring a dependency on slf4j-api matching the version of the declared provider/binding will make the problem go away.

Please also read the FAQ entry on backward compatibility for a more general explanation.

When using the fluent API, how can I make my IDE alert me when I forget to call the log() method?

When using the fluent API, you must terminate the invocation chain by calling one of the log() method variants. Forgetting to call any of the log() method variants will result in no logging regardless of the logging level. Fortunately, if this happens, some IDEs will alert you with a compiler warning.

IntelliJ IDEA will alert you if you activate the "Results of method call ignored" inspection. This inspection is configured Settings→Editor → Inspections → Java → Probable Bugs → "Results of method call ignored". This IDE inspection looks for methods annotated with "*.CheckReturnValue" annotations which the relevant methods in LoggingEventBuilder do include. Moreover, you probably want to set the severity of this inspection to "ERROR"

We have asked for a similar feature in Eclipse in Bug 572496. Please vote for this bug if you are also interested in it.

Why is SLF4J licensed under X11 type license instead of the Apache Software License?

SLF4J is licensed under a permissive X11 type license instead of the ASL or the LGPL because the X11 license is deemed by both the Apache Software Foundation as well as the Free Software Foundation as compatible with their respective licenses.

Where can I get a particular SLF4J provider/binding?

SLF4J providers such as NOPServiceProvider, SimpleLoggerProvider, Reload4jServiceProvider and JULServiceProvider are contained respectively within the files slf4j-nop.jar, slf4j-simple.jar, slf4j-reload4j.jar, and slf4j-jdk14.jar. These files can be found on Maven central. Please note that all providers/bindings depend on slf4j-api.jar.

The providers for logback-classic is part of the logback project and can also be found on Maven central. However, as with all other providers, the logback-classic provider requires slf4j-api.jar.

Should my library attempt to configure logging?

Embedded components such as libraries not only do not need to configure the underlying logging framework, they really should not do so. They should invoke SLF4J to log but should let the end-user configure the logging environment. When embedded components try to configure logging on their own, they often override the end-user's wishes. At the end of the day, it is the end-user who has to read the logs and process them. She should be the person to decide how she wants her logging configured.

In order to reduce the number of dependencies of our software we would like to make SLF4J an optional dependency. Is that a good idea?

This question pops up whenever a software project reaches a point where it needs to devise a logging strategy.

Let Wombat be a software library with very few dependencies. If SLF4J is chosen as Wombat's logging API, then a new dependency on slf4j-api.jar will be added to Wombat's list of dependencies. Given that writing a logging wrapper does not seem that hard, some developers will be tempted to wrap SLF4J and link with it only if it is already present on the classpath, making SLF4J an optional dependency of Wombat. In addition to solving the dependency problem, the wrapper will isolate Wombat from SLF4J's API ensuring that logging in Wombat is future-proof.

On the other hand, any SLF4J-wrapper by definition depends on SLF4J. It is bound to have the same general API. If in the future a new and significantly different logging API comes along, code that uses the wrapper will be equally difficult to migrate to the new API as code that used SLF4J directly. Thus, the wrapper is not likely to future-proof your code, but to make it more complex by adding an additional indirection on top of SLF4J, which is an indirection in itself.

increased vulnerability It is actually worse than that. Wrappers will need to depend on certain internal SLF4J interfaces which change from time to time, contrary to the client-facing API which never changes. Thus, wrappers are usually dependent on the major version they were compiled with. A wrapper compiled against SLF4J version 1.5.x will not work with SLF4J 1.6 whereas client code using org.slf4j.Logger, LoggerFactory, MarkerFactory, org.slf4j.Marker, and MDC will work fine with any SLF4J version from version 1.0 and onwards.

It is reasonable to assume that in most projects Wombat will be one dependency among many. If each library had its own logging wrapper, then each wrapper would presumably need to be configured separately. Thus, instead of having to deal with one logging framework, namely SLF4J, the user of Wombat would have to detail with Wombat's logging wrapper as well. The problem will be compounded by each framework that comes up with its own wrapper in order to make SLF4J optional. (Configuring or dealing with the intricacies of five different logging wrappers is not exactly exciting nor endearing.)

The logging strategy adopted by the Velocity project is a good example of the "custom logging abstraction" anti-pattern. By adopting an independent logging abstraction strategy, Velocity developers have made life harder for themselves, but more importantly, they made life harder for their users. Note that Velocity has since adopted slf4j.

Some projects try to detect the presence of SLF4J on the class path and switch to it if present. While this approach seems transparent enough, it will result in erroneous location information. Underlying logging frameworks will print the location (class name and line number) of the wrapper instead of the real caller. Then there is the question of API coverage as SLF4J support MDC and markers in addition to parameterized logging. While one can come up with a seemingly working SLF4J-wrapper within hours, many technical issues will emerge over time which Wombat developers will have to deal with. Note that SLF4J has evolved over several years and has 580 bug reports filed against it.

For the above reasons, developers of frameworks should resist the temptation to write their own logging wrapper. Not only is it a waste of time of the developer, it will actually make life more difficult for the users of said frameworks and make logging code paradoxically more vulnerable to change.

What about Maven transitive dependencies?

As an author of a library built with Maven, you might want to test your application using a provider, say slf4j-reload4j or logback-classic, without forcing reload4j or logback-classic as a dependency upon your users. This is rather easy to accomplish.

Given that your library's code depends on the SLF4J API, you will need to declare slf4j-api as a compile-time (default scope) dependency.

<dependency>
  <groupId>org.slf4j</groupId>
  <artifactId>slf4j-api</artifactId>
  <version>2.0.12</version>
</dependency>

Limiting the transitivity of the SLF4J provider used in your tests can be accomplished by declaring the scope of the SLF4J-provider dependency as "test". Here is an example:

<dependency>
  <groupId>org.slf4j</groupId>
  <artifactId>slf4j-reload4j</artifactId>
  <version>2.0.12</version>
  <scope>test</scope>
</dependency>

Thus, as far as your users are concerned you are exporting slf4j-api as a transitive dependency of your library, but not any SLF4J-provider or any underlying logging system.

Note that as of SLF4J version 1.6, in the absence of an SLF4J provider, slf4j-api will default to a no-operation implementation.

How do I exclude commons-logging as a Maven dependency?

alternative 1) explicit exclusion

Many software projects using Maven declare commons-logging as a dependency. Therefore, if you wish to migrate to SLF4J or use jcl-over-slf4j, you would need to exclude commons-logging in all of your project's dependencies which transitively depend on commons-logging. Dependency exclusion is described in the Maven documentation. Excluding commons-logging explicitly for multiple dependencies distributed on several pom.xml files can be a cumbersome and a relatively error-prone process.

alternative 2) provided scope

Commons-logging can be rather simply and conveniently excluded as a dependency by declaring it in the provided scope within the pom.xml file of your project. The actual commons-logging classes would be provided by jcl-over-slf4j. This translates into the following pom file snippet:

<dependency>
  <groupId>commons-logging</groupId>
  <artifactId>commons-logging</artifactId>
  <version>1.1.1</version>
  <scope>provided</scope>
</dependency>

<dependency>
  <groupId>org.slf4j</groupId>
  <artifactId>jcl-over-slf4j</artifactId>
  <version>2.0.12</version>
</dependency>

The first dependency declaration essentially states that commons-logging will be "somehow" provided by your environment. The second declaration includes jcl-over-slf4j into your project. As jcl-over-slf4j is a perfect binary-compatible replacement for commons-logging, the first assertion becomes true.

Unfortunately, while declaring commons-logging in the provided scope gets the job done, your IDE, e.g. Eclipse, will still place commons-logging.jar on your project's class path as seen by your IDE. You would need to make sure that jcl-over-slf4j.jar is visible before commons-logging.jar by your IDE.

alternative 3) empty artifacts

An alternative approach is to depend on an empty commons-logging.jar artifact. This clever approach first was imagined and initially supported by Erik van Oosten.

Such empty artifacts are available at a https://version99.qos.ch.

What are the JPMS module names of the various SLF4J artifacts?

Although compatible with Java 8, SLF4J version 2.0 supports JPMS modularisation as introduced in Java 9. Here are the JPMS module names for the various artifacts shipping in SLF4J.

artifact name JPMS module name
slf4j-api.jar org.slf4j
slf4j-simple.jar org.slf4j.simple
slf4j-jdk14.jar org.slf4j.jul
slf4j-nop.jar org.slf4j.nop
jcl-over-slf4j.jar org.apache.commons.logging
log4j-over-slf4j.jar log4j
slf4j-jdk-platform-logging.jar org.slf4j.jdk.platform.logging

What is the difference between an SLF4J binding and an SLF4J provider?

An SLF4J binding and a provider designate the same functionality, namely the act of tying the SLF4J API with its implementation at runtime.

In SLF4J 1.7.x and earlier this was done by invoking methods of StaticLoggerBinder class as found on the class path during LoggerFactory initialization. An artifact containing a StaticLoggerBinder class is called a "binding".

In SLF4J 2.0.x and later, during its initalization, LoggerFactory looks up SLF4JServiceProvider instances using the Java platfom's ServiceLoader mechanism. An artifact containing an implementation of the SLF4JServiceProvider interface is called a "provider".

About the SLF4J API

Why don't the printing methods in the Logger interface accept message of type Object, but only messages of type String?

In SLF4J 1.0beta4, the printing methods such as debug(), info(), warn(), error() in the Logger interface were modified so as to accept only messages of type String instead of Object.

Thus, the set of printing methods for the DEBUG level became:

debug(String msg); 
debug(String format, Object arg); 
debug(String format, Object arg1, Object arg2);           
debug(String msg, Throwable t);

Previously, the first argument in the above methods was of type Object.

This change enforces the notion that logging systems are about decorating and handling messages of type String, and not any arbitrary type (Object).

Just as importantly, the new set of method signatures offer a clearer differentiation between the overloaded methods whereas previously the choice of the invoked method due to Java overloading rules were not always easy to follow.

It was also easy to make mistakes. For example, previously it was legal to write:

logger.debug(new Exception("some error"));

Unfortunately, the above call did not print the stack trace of the exception. Thus, a potentially crucial piece of information could be lost. When the first parameter is restricted to be of type String, then only the method

debug(String msg, Throwable t);

can be used to log exceptions. Note that this method ensures that every logged exception is accompanied by a descriptive message.

Can I log an exception without an accompanying message?

In short, no.

If e is an Exception, and you would like to log an exception at the ERROR level, you must add an accompanying message. For example,

logger.error("some accompanying message", e);

You might legitimately argue that not all exceptions have a meaningful message to accompany them. Moreover, a good exception should already contain a self-explanatory description. The accompanying message may therefore be considered redundant.

While these are valid arguments, there are three opposing arguments also worth considering. First, on many, albeit not all occasions, the accompanying message can convey useful information nicely complementing the description contained in the exception. Frequently, at the point where the exception is logged, the developer has access to more contextual information than at the point where the exception is thrown. Second, it is not difficult to imagine more or less generic messages, e.g. "Exception caught", "Exception follows", that can be used as the first argument for error(String msg, Throwable t) invocations. Third, most log output formats display the message on a line, followed by the exception on a separate line. Thus, the message line would look inconsistent without a message.

In short, if the user were allowed to log an exception without an accompanying message, it would be the job of the logging system to invent a message. This is actually what the throwing(String sourceClass, String sourceMethod, Throwable thrown) method in java.util.logging package does. (It decides on its own that accompanying message is the string "THROW".)

It may initially appear strange to require an accompanying message to log an exception. Nevertheless, this is common practice in all log4j derived systems such as java.util.logging, logkit, etc. and of course log4j itself. It seems that the current consensus considers requiring an accompanying message as a good a thing (TM).

What is the fastest way of (not) logging?

SLF4J supports a feature called parameterized logging which can significantly boost logging performance for disabled logging statements.

For some Logger logger, writing,

logger.debug("Entry number: " + i + " is " + String.valueOf(entry[i]));

incurs the cost of constructing the message parameter, that is converting both integer i and entry[i] to a String, and concatenating intermediate strings. This, regardless of whether the message will be logged or not.

One possible way to avoid the cost of parameter construction is by surrounding the log statement with a test. Here is an example.

if(logger.isDebugEnabled()) {
  logger.debug("Entry number: " + i + " is " + String.valueOf(entry[i]));
}

This way you will not incur the cost of parameter construction if debugging is disabled for logger. On the other hand, if the logger is enabled for the DEBUG level, you will incur the cost of evaluating whether the logger is enabled or not, twice: once in debugEnabled and once in debug. This is an insignificant overhead because evaluating a logger takes less than 1% of the time it takes to actually log a statement.

Better yet, use parameterized messages

There exists a very convenient alternative based on message formats. Assuming entry is an object, you can write:

Object entry = new SomeObject();
logger.debug("The entry is {}.", entry);

After evaluating whether to log or not, and only if the decision is affirmative, will the logger implementation format the message and replace the '{}' pair with the string value of entry. In other words, this form does not incur the cost of parameter construction in case the log statement is disabled.

The following two lines will yield the exact same output. However, the second form will outperform the first form by a factor of at least 30, in case of a disabled logging statement.

logger.debug("The new entry is "+entry+".");
logger.debug("The new entry is {}.", entry);

A two argument variant is also available. For example, you can write:

logger.debug("The new entry is {}. It replaces {}.", entry, oldEntry);

If three or more arguments need to be passed, you can make use of the Object... variant of the printing methods. For example, you can write:

logger.debug("Value {} was inserted between {} and {}.", newVal, below, above);

This form incurs the hidden cost of construction of an Object[] (object array) which is usually very small. The one and two argument variants do not incur this hidden cost and exist solely for this reason (efficiency). The slf4j-api would be smaller/cleaner with only the Object... variant.

Array type arguments, including multidimensional arrays, are also supported.

SLF4J uses its own message formatting implementation which differs from that of the Java platform. This is justified by the fact that SLF4J's implementation performs about 10 times faster but at the cost of being non-standard and less flexible.

Escaping the "{}" pair

The "{}" pair is called the formatting anchor. It serves to designate the location where arguments need to be substituted within the message pattern.

SLF4J only cares about the formatting anchor, that is the '{' character immediately followed by '}'. Thus, in case your message contains the '{' or the '}' character, you do not have to do anything special unless the '}' character immediately follows '{'. For example,

logger.debug("Set {1,2} differs from {}", "3");

which will print as "Set {1,2} differs from 3".

You could have even written,

logger.debug("Set {1,2} differs from {{}}", "3");

which would have printed as "Set {1,2} differs from {3}".

In the extremely rare case where the "{}" pair occurs naturally within your text and you wish to disable the special meaning of the formatting anchor, then you need to escape the '{' character with '\', that is the backslash character. Only the '{' character should be escaped. There is no need to escape the '}' character. For example,

logger.debug("Set \\{} differs from {}", "3");

will print as "Set {} differs from 3". Note that within Java code, the backslash character needs to be written as '\\'.

In the rare case where the "\{}" occurs naturally in the message, you can double escape the formatting anchor so that it retains its original meaning. For example,

logger.debug("File name is C:\\\\{}.", "file.zip");

will print as "File name is C:\file.zip".

How can I log the string contents of a single (possibly complex) object?

In relatively rare cases where the message to be logged is the string form of an object, then the parameterized printing method of the appropriate level can be used. Assuming complexObject is an object of certain complexity, for a log statement of level DEBUG, you can write:

logger.debug("{}", complexObject);

The logging system will invoke complexObject.toString() method only after it has ascertained that the log statement was enabled. Otherwise, the cost of complexObject.toString() conversion will be advantageously avoided.

Why doesn't the org.slf4j.Logger interface have methods for the FATAL level?

The Marker interface, part of the org.slf4j package, renders the FATAL level largely redundant. If a given error requires attention beyond that allocated for ordinary errors, simply mark the logging statement with a specially designated marker which can be named "FATAL" or any other name to your liking.

Here is an example,

import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.slf4j.Marker;
import org.slf4j.MarkerFactory;

class Bar {
  void foo() {
    Marker fatal = MarkerFactory.getMarker("FATAL");
    Logger logger = LoggerFactory.getLogger("aLogger");

    try {
      ... obtain a JDBC connection
    } catch (JDBException e) {
      logger.error(fatal, "Failed to obtain JDBC connection", e);
    }
  }
}

While markers are part of the SLF4J API, only logback supports markers off the shelf. For example, if you add the %marker conversion word to its pattern, logback's PatternLayout will add marker data to its output. Marker data can be used to filter messages or even trigger an outgoing email at the end of an individual transaction.

In combination with logging frameworks such as log4j and java.util.logging which do not support markers, marker data will be silently ignored.

Markers add a new dimension with infinite possible values for processing log statements compared to five values, namely ERROR, WARN, INFO, DEBUG and TRACE, allowed by levels. At present time, only logback supports marker data. However, nothing prevents other logging frameworks from making use of marker data.

Why was the TRACE level introduced only in SLF4J version 1.4.0?

The addition of the TRACE level has been frequently and hotly debated request. By studying various projects, we observed that the TRACE level was used to disable logging output from certain classes without needing to configure logging for those classes. Indeed, the TRACE level is by default disabled in log4j and logback as well most other logging systems. The same result can be achieved by adding the appropriate directives in configuration files.

Thus, in many of cases the TRACE level carried the same semantic meaning as DEBUG. In such cases, the TRACE level merely saves a few configuration directives. In other, more interesting occasions, where TRACE carries a different meaning than DEBUG, Marker objects can be put to use to convey the desired meaning. However, if you can't be bothered with markers and wish to use a logging level lower than DEBUG, the TRACE level can get the job done.

Note that while the cost of evaluating a disabled log request is in the order of a few nanoseconds, the use of the TRACE level (or any other level for that matter) is discouraged in tight loops where the log request might be evaluated millions of times. If the log request is enabled, then it will overwhelm the target destination with massive output. If the request is disabled, it will waste resources.

In short, although we still discourage the use of the TRACE level because alternatives exist or because in many cases log requests of level TRACE are wasteful, given that people kept asking for it, we decided to bow to popular demand.

What are markers and what are they goood for?

Markers can be used to color or mark a single log statement. What you do with these colors, i.e. markers, is entirely up to you. However, two patterns seem to be common for marker usage.

  1. Triggering: Some appender could be instructed to take an action in the presence of a certain marker. For example, SMTPAppender can be configured to send an email whenever a logging event is marked with the NOTIFY_ADMIN marker regardless of the log level. See marker-based triggering in the logback documentation. You may also combine log levels and markers for triggering.
  2. Filtering: Markers are very useful for making certain valuable log statements stand out. For example, you can color/mark all your persistence related logs (in various and multiple class files) with the color "DB". You could then filter for "DB": to disable logging except for log statements marked with DB. See the chapter on filters in the logback documentation for more information (search for MarkerFilter). Note that filtering on markers can be performed not just by logback but by log analysis tools as well.

Before the advent of Markers, to achieve similar behavior, you had the option 1) using custom levels 2) use modified logger names. SLF4J API currently does not support custom levels. As for option 2, suffixing (or prefixing) logger names is workable if a one or two loggers need to be modified. The approach becomes impractical as soon 3 or more loggers need to be "sub-classed" because the associated configuration files become unmanageable.

Even though a single marker can be already very useful, as of SLF4J version 2.0, it is possible to set multiple markers per log statement.

Does the SLF4J logging API support I18N (internationalization)?

Yes, as of version 1.5.9, SLF4J ships with a package called org.slf4j.cal10n which adds localized/internationalized logging support as a thin layer built upon the CAL10N API.

Is it possible to retrieve loggers without going through the static methods in LoggerFactory?

Yes. LoggerFactory is essentially a wrapper around an ILoggerFactory instance. The ILoggerFactory instance in use is determined according to the static provider conventions of the SLF4J framework. See the getSingleton() method in LoggerFactory for details.

However, nothing prevents you from using your own ILoggerFactory instance. Note that you can also obtain a reference to the ILoggerFactory that the LoggerFactory class is using by invoking the LoggerFactory.getILoggerFactory() method.

Thus, if SLF4J binding conventions do not fit your needs, or if you need additional flexibility, then do consider using the ILoggerFactory interface as an alternative to inventing your own logging API.

In the presence of an exception/throwable, is it possible to parameterize a logging statement?

Yes, as of SLF4J 1.6.0, but not in previous versions. The SLF4J API supports parametrization (of 1 or more objects) in the presence of an exception, assuming the exception is the last parameter. Thus,

String s = "Hello world";
try {
  Integer i = Integer.valueOf(s);
} catch (NumberFormatException e) {
  logger.error("Failed to format {}", s, e);
}

will print the NumberFormatException with its stack trace as expected. The java compiler will invoke the error method taking a String and two Object arguments. SLF4J, in accordance with the programmer's most probable intention, will interpret NumberFormatException instance as a throwable instead of an unused Object parameter. In SLF4J versions prior to 1.6.0, the NumberFormatException instance was simply ignored.

If the exception is not the last argument, it will be treated as a plain object and its stack trace will NOT be printed. However, such situations should not occur in practice.

As mentioned earlier, it is possible to pass one or more parameters in addition to the exception.

Implementing the SLF4J API

How do I make my logging framework SLF4J compatible?

Adding supporting for the SLF4J is surprisingly easy. Essentially, you copying an existing provider and tailoring it a little (as explained below) should do the trick.

Assuming your logging system has notion of a logger, called say MyLogger, you need to provide an adapter for MyLogger to org.slf4j.Logger interface. Refer to slf4j-jcl, slf4j-jdk14, and slf4j-reload4j modules for examples of adapters.

Instead of adapting your existing Logger to the org.slf4j.Logger interface, you may also implement the org.slf4j.Logger interface directly, in which case no adapter would be necessary. This is the approach taken by the slf4j-simple module.

Once you have written an appropriate adapter or logger implementation, say MyLoggerAdapter, you need to provide a factory class implementing the org.slf4j.ILoggerFactory interface. This factory must return instances MyLoggerAdapter. Let MyLoggerFactory be the name of your factory class.

The last remaining step is to create a provider class say MySLF4JServiceProvider implementing the org.slf4j.spi.SLF4JServiceProvider interface. The MySLF4JServiceProvider class must be declared in the META-INF/services/org.slf4j.spi.SLF4JServiceProvider file of your module.

For Marker or MDC support, you can use the one of the existing implementations or write your own extensions.

In summary, to create an SLF4J provider for your logging system, follow these steps:

  1. start with a copy of an existing module,
  2. create an adapter between your logging system and org.slf4j.Logger interface
  3. create a factory for the adapter created in the previous step,
  4. implement SLF4JServiceProvider class to use the factory you created in the previous step
  5. declare your provider in META-INF/services/org.slf4j.spi.SLF4JServiceProvider

How can my logging system add support for the Marker interface?

Markers are a relatively new feature which are supported several but not all logging systems. Consequently, SLF4J conforming logging systems are allowed to ignore marker data passed by the user.

However, even though marker data may be ignored, the user must still be allowed to specify marker data. Otherwise, users would not be able to switch between logging systems that support markers and those that do not.

The MarkerIgnoringBase class can serve as a base for adapters or native implementations of logging systems lacking marker support. In MarkerIgnoringBase, methods taking marker data simply invoke the corresponding method without the Marker argument, discarding any Marker data passed as argument. Your SLF4J adapters can extend MarkerIgnoringBase to quickly implement the methods in org.slf4j.Logger which take a Marker as the first argument.

How does SLF4J's version check mechanism work?

As of version 2.0.x, the version check performed by SLF4J API during initialization is a mandatory process. Conforming SLF4J implementations must return the desired API version via the getRequestedApiVersion method in the class implementing the SLF4JServiceProvider interface.

For each version, SLF4J API maintains a list of compatible versions. SLF4J will emit a version mismatch warning only if the requested version is not found in the compatibility list. So even if your SLF4J provider has a different release schedule than SLF4J, assuming you update the SLF4J version you use every 6 to 12 months, you can still participate in the version check without incurring a mismatch warning. For example, logback has a different release schedule but still participates in version checks.

General questions about logging

Should Logger members of a class be declared as static?

We used to recommend that loggers members be declared as instance variables instead of static. After further analysis, we no longer recommend one approach over the other.

Here is a summary of the pros and cons of each approach.

Advantages for declaring loggers as static Disadvantages for declaring loggers as static
  1. common and well-established idiom
  2. less CPU overhead: loggers are retrieved and assigned only once, at hosting class initialization
  3. less memory overhead: logger declaration will consume one reference per class
  1. For libraries shared between applications, not possible to take advantage of repository selectors. It should be noted that if the SLF4J provider and the underlying API ships with each application (not shared between applications), then each application will still have its own logging environment.
  2. not IOC-friendly
Advantages for declaring loggers as instance variables Disadvantages for declaring loggers as instance variables
  1. Possible to take advantage of repository selectors even for libraries shared between applications. However, repository selectors only work if the underlying logging system is logback-classic. Repository selectors do not work for the SLF4J+log4j combination.
  2. IOC-friendly
  1. Less common idiom than declaring loggers as static variables
  2. higher CPU overhead: loggers are retrieved and assigned for each instance of the hosting class
  3. higher memory overhead: logger declaration will consume one reference per instance of the hosting class

Explanation

Static logger members cost a single variable reference for all instances of the class whereas an instance logger member will cost a variable reference for every instance of the class. For simple classes instantiated thousands of times there might be a noticeable difference.

However, more recent logging systems, e.g. log4j or logback, support a distinct logger context for each application running in the application server. Thus, even if a single copy of log4j.jar or logback-classic.jar is deployed in the server, the logging system will be able to differentiate between applications and offer a distinct logging environment for each application.

More specifically, each time a logger is retrieved by invoking LoggerFactory.getLogger() method, the underlying logging system will return an instance appropriate for the current application. Please note that within the same application retrieving a logger by a given name will always return the same logger. For a given name, a different logger will be returned only for different applications.

If the logger is static, then it will only be retrieved once when the hosting class is loaded into memory. If the hosting class is used in only in one application, there is not much to be concerned about. However, if the hosting class is shared between several applications, then all instances of the shared class will log into the context of the application which happened to first load the shared class into memory - hardly the behavior expected by the user.

Unfortunately, for non-native implementations of the SLF4J API, namely with slf4j-log4j12, log4j's repository selector will not be able to do its job properly because slf4j-log4j12, a non-native SLF4J provider, will store logger instances in a map, short-circuiting context-dependent logger retrieval. For native SLF4J implementations, such as logback-classic, repository selectors will work as expected.

The Apache Commons wiki contains an informative article covering the same question.

Logger serialization

Contrary to static variables, instance variables are serialized by default. As of SLF4J version 1.5.3, logger instances survive serialization. Thus, serialization of the host class no longer requires any special action, even when loggers are declared as instance variables. In previous versions, logger instances needed to be declared as transient in the host class.

Summary

In summary, declaring logger members as static variables requires less CPU time and have a slightly smaller memory footprint. On the other hand, declaring logger members as instance variables requires more CPU time and have a slightly higher memory overhead. However, instance variables make it possible to create a distinct logger environment for each application, even for loggers declared in shared libraries. Perhaps more important than previously mentioned considerations, instance variables are IOC-friendly whereas static variables are not.

See also related discussion in the commons-logging wiki.

Is there a recommended idiom for declaring a logger in a class?

The following is the recommended logger declaration idiom. For reasons explained above, it is left to the user to determine whether loggers are declared as static variables or not.

package some.package;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
      
public class MyClass {
  final (static) Logger logger = LoggerFactory.getLogger(MyClass.class);
  ... etc
}

Unfortunately, given that the name of the hosting class is part of the logger declaration, the above logger declaration idiom is not resistant to cut-and-pasting between classes.

Alternatively, you can use MethodHandles.lookup() introduced in JDK 7 to pass the caller class.

package some.package;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import java.lang.invoke.MethodHandles;
      
public class MyClass {
  final static Logger logger = LoggerFactory.getLogger(MethodHandles.lookup().lookupClass());
  ... etc
}

This pattern can be cut and pasted across classes.