JVM only

Use Java8/Scala (/possibly Clojure)

Use functional constructs where possible (eg beans should be immutable - see below)

Use guice for dependency injection of shared services

Access Layer applications should use the JAX-RS standard and will likely run in Tomcat7/8 (we are also investigating the play framework). They should use guice (the data_model will provide a utility for injecting the Aleph2 services) via guice-servlet.

null should be avoided - use Optional instead. Use the @NonNull annotation where possible.

Futures should be used where there is any chance of concurrency.

All database objects should be representing as "pojos" (referred henceforth, not quite accurately, as "beans"):

• default constructor (plus user constructor)
• all fields private
• getters and setters should be in the format "<fieldname>()"/"<fieldname>(T t)" ** (actually setters are discouraged since beans should be immutable, see below)

In addition:

• immutable (no setters - use data_model.util.ObjectUtils to clone; all collections should be passed through collection.unmodifiableXXX) (more work is needed on handling collection-of-collections immutability)
• minimize within a bean nesting
• fields should be "JSON-izable", eg lists, sets, and linkedhashmaps

Jackson should be used for serialization/deserialization.

In interfaces and beans, all function names should be "Drinking Camel Case" eg drinkingCamelCase and all variable names should be "lower case and underscored", eg 'lower_case_and_underscored'

When accessing fields from beans, ObjectUtils should be used together with getter syntax (eg XxxBean::fieldname) vs raw strings ("fieldname"). This is currently not supported for nested fields or maps, so they will have to remain as strings for now.

Projects should be called aleph2_<function>_<technology>, eg aleph2_archive_layer_hdfs, aleph2_search_index_layer_elasticsearch etc.

(We may used github submodules for Aleph2 and Aleph2-contrib in the future, to enable per project CI.)

JUnit and Mockito should be used for testing

In eclipse ecl-emma should be used for code coverage

In github, travis and coveralls (JoCoCo) should be used for continuous integration