blob: 0149504f36bb0af58ad58ca0138cd418370ef1cc [file] [log] [blame]
:jbake-type: page
:jbake-status: published
toc::[]
= Apache Tamaya - Possible Tasks
:numbered:
<<<
== Introduction
== What is Tamaya
{name} is the Apache standard for flexible and powerful configuration. Objective is to provide flavors for
Java SE, ME as well as to ship with powerful features for Java EE and Cloud Solutions. All functions provided
is build on top of a small but very powerful, flexible and extendible API. This API is implemented by a core implementation,
which then can be extended or adapted for use in different runtime scenarios, such as SE, ME, EE, Spring, OSGi
and more. Similarly additional modules may be provided that help also existing solution to be plugged into
{name}, so you can start right away using {name} without having to rebuild/change your existing application.
=== Purpose of this Document
The document should help to organize people and ideas around the Apache Tamaya Library. It list possible features,
ideas and tasks that need to be done. Everybody can have a look at and see, where hos contribution and capabilities
would fit best.
== Main Features
=== Metadata Model
Currently +MetaInfo+ models metadata as a separate constuct. It has been shown that this leads to more complex
handling when creating composites and makes the API overall more complex. The idea is to model metadata as simple
key/value pairs, that are part of the provider/configuration data as well, but handled separately. Metadata hereby
is identified by a starting '_' character in its key. For example refer to the following configuration properties:
[source,listing]
.Basic Properties
----------------------------------------------------------------
a.b.Foo=foo
a.b.Bar=bar
a.AnyOther=whatelse
Something=none
----------------------------------------------------------------
Now we can model meta-data as follows:
[source,listing]
.Metadata Properties
----------------------------------------------------------------
[a.b].info=An area info
[a.b.Foo].auth=role1,role2
[a.b.Foo].encrypt=PGP
[a.b.Foo].sensitive=true
[].info=This is a test configuration example.
----------------------------------------------------------------
The above would model the following:
* The area +a.b+ has the meta property +info+.
* The entry +a.b.Foo+ has three meta properties +auth,encrypt+ and +sensitive+. These could be interpreted by a security
view and used to encrypt the values returned by the configuration instance, if not the current user has one of the
specified roles.
* The last meta data defines an attribute +info+ for the whole provider/configuration (the root area).
Given that the overall entries would be as follows:
[source,listing]
.Full Properties with Meta Properties
----------------------------------------------------------------
[a.b].info=An area info
a.b.Foo=foo
[a.b.Foo].auth=role1,role2
[a.b.Foo].encrypt=PGP
[a.b.Foo].sensitive=true
a.b.Bar=bar
[].info=This is a test configuration example.
a.AnyOther=whatelse
Something=none
----------------------------------------------------------------
The current +MetaInfo+ class could be adapted, so it is reading data from the underlying configuration/provider,
instead of its own datastructure. This would make a later mapping of configuration and its metadata into DB table, JSON
etc, much more easier.
The providers on the other side may suppress any metadata from ordinary output, such
as +toString()+, Similarly accessing metadata using the official config API (+get, getOrDefault, getAreas+ etc)
should be disabled. The +MetaInfoBuilder+ must probably as well adapted or redesigned.
=== Management Client
A nice web-based client to manage configuration data would be nice as well. This also includes a UI for creating new
configurations.
=== Mapping Configuration to a Database
A flexible mechanism should be implemented that allows the use of databases (SQL/JPA as well as non-SQL) for
storing/retreiving/managing configuration:
* JPA, Hibernate
* MongoDB
* ...
=== Integration with Jigsaw
Once Jigsaw is mature and in a usable (still early) stage, examples are to be created and tested, where OSGi is used as
the basic runtime platform, e.g. Apache Felix, but as well others.
== Distributed/Remote Configuration Support
=== Configuration Distribution Policies
Different configuration distribution policies should be defined any implemented, e.g. distributed cache, restful services,
web services, EJB/RMI calls, asynchronous queues, publish/subsribe models, ...
=== Preferences Support
Write a +PreferencesFactory+ for +java.util.preferences+.
== Third Party Integration
=== Integration with Deltaspike Config
Integration with Deltaspike Config should be implemented and discussed with Deltaspike guys.
=== Integration with Spring
A {name} module should be created that allows Spring to be used either as client or configuration provider.
=== Integration with Jetty
A {name} module should be created that allows a Jetty instance to be deployed and started that is (completely)
configured based on configuration server.
=== Integration with Tomcat
A {name} module should be created that allows a Tomcat instance to be deployed and started that is (completely)
configured based on configuration server.
=== Configuration of Java EE
In the Java EE area there would be several options:
=== Configuration of Application Servers (administrative resources)
It should be possible to start a application server instance remotely and configure all administrative resources and the
deployments based on the configuration service, server to be considered maybe
* Wildfly
* IBM
* Weblogic
* Glassfish
* Apache Geronimo
==== Configuration of Bean Validation
* Add configurable validators.
* Configure bean validation based on configuration
* ...
=== JNDI Support
Write a +JCA+ adapter to provide configuration data through JNDI.
==== Configure JSF
Use the JSF +XML Document+ event to completely configure JSF.
==== Configure Web Services
Provide a WebServiceProviderFactory that may be configured.
==== Configure JPA
Provide an implementation that allows configuration of persistence units. Talk with JPA EG people to see if we can
get an SPI to hook in a stadardized way.
==== Configure EJBs
Provide an implementation that allows configuration of EJBs and MDBs:
* Register beans
* Unregister/disable beans
* Intercept beans
* Support Configuration Injection (in the worst case using a standard Interceptor, provide supporting artifacts to
help developers to achive this easily).
* Talk with EE8 Umbrella EG (Bill Shanon, Linda DeMichels) on a feasible SPI for EE8, if possible join the EG.
==== Configure ...
Just think of any Java EE aspects that might be worth to be configured. If it can be done, e.g. by managing CDI managed
resources, it might be easy. For others it is a good idea to discuss things with our matter of experts...
== Special Goodies
=== Maintenance Mode Servlet Filter
Provide a servlet filter that is capable of switching to maintenance mode, based on configuration. Similarly also a forwarding
servlet could be useful, wehere only request based on configuration are forwarded, other might be rejected or dropped
as configured.
=== Dynamic Camel Routes
Provides dynamic (configurable) Camel routes, e.g. usable within ServiceMix or standalone.
=== Dynamic CXF
Provides dynamic (configurable) CXF adapters, e.g. usable within ServiceMix or standalone.
=== Configurable Apache MQ
Provides an implementation for configuring Apache MQ.
=== Dynamic ...
Interested to see what other ideas are around. Let us know!