Javaslang needs to be compiled with jdk 1.8.0_40 at a minimum, which fixes many type inference bugs of the java compiler.
Fork the GitHub, send a pull request and keep your fork in sync with the upstream repository.
We use recent IDE version to develop Javaslang. IntelliJ IDEA is preferred over Eclipse.
Using IntelliJ IDEA, the Community Edition works out-of-the-box. The idea-settings.jar can be found in the repo.
The IDEA settings include:
- Code Style
- File templates
- Inspection profiles
We follow Rob Pike's 5 Rules of Programming:
- Rule 1. You can't tell where a program is going to spend its time. Bottlenecks occur in surprising places, so don't try to second guess and put in a speed hack until you've proven that's where the bottleneck is.
- Rule 2. Measure. Don't tune for speed until you've measured, and even then don't unless one part of the code overwhelms the rest.
- Rule 3. Fancy algorithms are slow when n is small, and n is usually small. Fancy algorithms have big constants. Until you know that n is frequently going to be big, don't get fancy. (Even if n does get big, use Rule 2 first.)
- Rule 4. Fancy algorithms are buggier than simple ones, and they're much harder to implement. Use simple algorithms as well as simple data structures.
- Rule 5. Data dominates. If you've chosen the right data structures and organized things well, the algorithms will almost always be self-evident. Data structures, not algorithms, are central to programming.
Pike's rules 1 and 2 restate Tony Hoare's famous maxim "Premature optimization is the root of all evil." Ken Thompson rephrased Pike's rules 3 and 4 as "When in doubt, use brute force.". Rules 3 and 4 are instances of the design philosophy KISS. Rule 5 was previously stated by Fred Brooks in The Mythical Man-Month. Rule 5 is often shortened to "write stupid code that uses smart objects".
Source: http://users.ece.utexas.edu/~adnan/pike.html
- Public API needs javadoc, e.g. public classes and public methods.
- Non-trivial private methods need javadoc, too.
- A package, which is part of the public API, contains a
package-info.java
. - Unit tests contain no javadoc at all (because they introduce no new API and contain no business logic).
- Running
mvn javadoc:javadoc
results in no javadoc errors. - All classes start with the following copyright notice, which contains the list of core developers:
/* / \____ _ _ ____ ______ / \ ____ __ _______
* / / \/ \ / \/ \ / /\__\/ // \/ \ // /\__\ JΛVΛSLΛNG
* _/ / /\ \ \/ / /\ \\__\\ \ // /\ \ /\\/ \ /__\ \ Copyright 2014-2017 Javaslang, http://javaslang.io
* /___/\_/ \_/\____/\_/ \_/\__\/__/\__\_/ \_// \__/\_____/ Licensed under the Apache License, Version 2.0
*/
- There is only one first-level package: javaslang.
- The maximum package depth is two.
- Package names are denoted in the singular.
- Packages are sliced by domain (no util or tool packages).
- Package private classes are used in order to hide non-public API.
- Inner classes are preferred over package private classes in case of one-to-one dependencies.
We organize our classes and interfaces in the following way:
- The Javadoc of the type contains an overview of the new (i.e. not overridden) API declared in the actual type.
- The type consists of three sections:
- static API
- non-static API
- adjusted return types
- The methods of each of these sections are alphabetically ordered.
/**
* Description of this class.
*
* <ul>
* <li>{@link #containsKey(Object)}}</li>
* <li>{@link ...}</li>
* </ul>
*
* @author ...
* @since ...
*/
public interface Map<K, V> extends Traversable<Tuple2<K, V>> {
// -- static API
static <K, V> Tuple2<K, V> entry(K key, V value) { ... }
...
// -- non-static API
@Override
default boolean contains(Tuple2<K, V> element) { ... }
boolean containsKey(K key);
...
// -- Adjusted return types
@Override
Map<K, V> distinct();
...
}
- Public API is tested.
- High-level functionality is tested in first place.
- Corner cases are tested.
- Trivial methods are not directly tested, e.g. getters, setters.
- The test method name documents the test, i.e. 'shouldFooWhenBarGivenBaz'
- In most cases it makes sense to run one assertion per @Test.
- Javaslang has no dependencies other than Java.
- Unit tests depend solely on junit and assertj.
- Benchmarks are done with JMH
- Executing tests:
mvn clean test
- Executing doclint:
mvn javadoc:javadoc
- Executing code coverage report:
mvn -P ci clean test jacoco:report
- Create -javadoc.jar:
mvn javadoc:jar
- Create -source.jar:
mvn source:jar
- Update version properties:
mvn versions:update-properties
- Check for new plugin version:
mvn versions:display-plugin-updates
If you have dedicated hardware (i.e. no virtual machines) and are interested in how Javaslang compares to other alternatives,
you can run all benchmarks from the javaslang-benchmark
module via javaslang.JmhRunner.main
or running the following Maven command:
mvn clean test -P benchmark -pl javaslang-benchmark
Note: running all the tests will require several hours, during which there should be no other activity done on the given machine.
See http://central.sonatype.org/pages/ossrh-guide.html
Sonatype-Nexus specific maven configuration: ~/.m2/settings.xml
<settings>
<servers>
<server>
<id>sonatype-nexus-snapshots</id>
<username>your-jira-id</username>
<password>your-jira-pwd</password>
</server>
<server>
<id>sonatype-nexus-staging</id>
<username>your-jira-id</username>
<password>your-jira-pwd</password>
</server>
</servers>
</settings>
Note: Detailed information about performing a release can be found in the SCM section.
- Commits are coarsely granular grouped by feature/change.
- Commits do not mix change sets of different domains/purpose.
- Commit messages provide enough detail to extract a changelog for a new release.
We following a simple git workflow/branching model:
master
|
| v2.0.x
release v2.0.0 - - - - - - + - - - + 2.0.1-SNAPSHOT
| |
bugfix1 | |
| | |
PR x---->|<------+ cherry-picking bugfix1
| |
featureA | |
| | |
PR x---->| |
| |
release v2.0.1 - - - - - - | - - - + 2.0.2-SNAPSHOT
| |
| | v2.1.x
release v2.1.0 - - - - - - + - - - X - - - + 2.1.1-SNAPSHOT
| |
| |
featureB | |
| | |
PR x---->| |
... ...
We follow the Semantic Versioning scheme.
We distinguish between 3 kinds of (backwward-)compatibilty:
- Source - Source compatibility concerns translating Java source code into class files.
- Binary - Binary compatibility is defined in The Java Language Specification as preserving the ability to link without error.
- Behavioral - Behavioral compatibility includes the semantics of the code that is executed at runtime.
Source: OpenJDK Developers Guide v0.777, Kinds of Compatibility
We currently check for API changes (which may affect the binary compatibility) using the maven-bundle-plugin:
mvn package org.apache.felix:maven-bundle-plugin:baseline -DcomparisonVersion=2.0.1 -DskipTests
In the example above we check API changes between the current branch and the 2.0.1 tag. In most cases the tag should be the latest official release.
Performing a release requires admin-rights.
-
get a fresh copy of the repo
git clone https://github.com/javaslang/javaslang.git
-
run
mvn clean test
and(!)mvn javadoc:javadoc
to ensure all is working fine -
perform the release
mvn release:clean mvn release:prepare mvn release:perform
-
Go to
http://oss.sonatype.org
and stage the release.
-
[CAUTION] Delete the old maintenance branch (e.g. v2.0.x)
git push origin :v2.0.x
-
Create the new maintenance branch (e.g. v2.1.x) based on the new release tag (e.g. v2.1.0)
git checkout origin/master git fetch origin git branch v2.1.x v2.1.0 git checkout v2.1.x git push origin v2.1.x
-
Update the version of the maintenance branch
mvn versions:set -DnewVersion=2.1.1-SNAPSHOT
When a maintenance release is performed, we increase the last digit of the new development version of the maintenance branch (e.g. 2.1.2-SNAPSHOT).
Pull requests are merged into master. Only specific commits are merged from master into the maintenance branch.
git checkout v2.1.x
git log --date-order --date=iso --graph --full-history --all --pretty=format:'%x08%x09%C(red)%h %C(cyan)%ad%x08%x08%x08%x08%x08%x08%x08%x08%x08%x08%x08%x08%x08%x08%x08 %C(bold green)%aN%C(reset)%C(bold yellow)%d %C(reset)%s'
# pick one or more commits from the log, e.g. a741cf1.
git cherry-pick a741cf1
Given a release 1.2.2, we create a bugfix release as follows.
First, we clone the repository. We work on origin instead of a fork (this requires admin rights).
git clone https://github.com/javaslang/javaslang.git javaslang-1.2.3
We checkout the release tag and create a new (local) branch.
git checkout v1.2.2
git checkout -b bugfix-release-1.2.3
Then we create the new snapshot version.
mvn versions:set -DnewVersion=1.2.3-SNAPSHOT
Now the changes can be performed to the repository. After that, we test the changes.
mvn clean test
mvn javadoc:javadoc
Then the new files can be added and the changes can be committed.
git add <files>
git commit -a -m "fixes #<issue>"
Then we perform the release as usual:
mvn release:clean
mvn release:prepare
mvn release:perform
Goto oss.sonatype.org and release to Maven Central.
Delete the branch which was pushed by the maven release plugin to origin:
git checkout master
git branch -D bugfix-release-1.2.3
git push origin :bugfix-release-1.2.3
For major, minor and bugfix releases we create release notes on Github.
The number of lines changed by author since a specific date:
git ls-files -z | xargs -0n1 git blame -w --since="3/18/2016" --line-porcelain | grep -a "^author " | sort -f | uniq -c | sort -n -r