Skip to main content

Notice: this Wiki will be going read only early in 2024 and edits will no longer be possible. Please see: https://gitlab.eclipse.org/eclipsefdn/helpdesk/-/wikis/Wiki-shutdown-plan for the plan.

Jump to: navigation, search

Gyrex/Context Runtime

Introduction

Eclipse is a dynamic and extensible platform. For example, the Eclipse extension registry allows for a great extensibility where some code defines an extension point and allows others to provide an implementation for it. Another possibility are OSGi services.

The Gyrex Platform includes a context sensitive runtime to support dynamic constraints. Every operation in Gyrex happens within (or on behalf of) a particular context. The contextual runtime is one of the key elements of Gyrex and allows - for example - the development of multi-tenant software offerings.

Let's look at an example. In an online shopping system a shopping cart calculation could be made extensible using the extension registry or using OSGi services. A default implementation would be provided which does a simple total calculation. This will be the default for all shops served by your system. The other day a new client comes along and has very special requirements for shopping cart calculations. For example, it might operate in a different country with specialized tax rules, etc. With OSGi or the Eclipse extension registry this is a no-brainer. You simply implement the special calculation and provide it as an OSGi service. But now the "host" code needs to deal with two services (or two extension). It need to know which one to use in which shop. Here comes the contextual runtime to the rescue. Using the contextual runtime a shop will be defined as the runtime context. Your code simply delegates the call. Instead of asking the extension registry or the OSGi service registry directly, it asks the context which service (or extension) to use. The rest will all be configurable through an administration interface (or via APIs). Your code does not need to implement any filtering or permission checks. The contextual runtime takes care of that.


The Runtime Context

The runtime context is the central element of the contextual runtime. It provides APIs to get a specific, ready-configured service for the client code base to use. The lookup of the service will be very simple. You simply pass a class object to context and it will return an instance of the class for you to work with. Thus, the context is essentially an object registry. However, it's dynamic, i.e. it can change between invocations.

Contexts are hierarchical organized in a path like structure. There is a root context ("/") which typically defines the core pieces of a system. In a single system it's actually possible to just work with the root context.

Context values are inherited. Thus, if a context does not have a value defined, it's parent will be queried until a value is found. It's also possible to explicitly undefine (set to "null") a value in a client context.

Note, for security reasons (see below) a context will not allow simple retrieval of its parent context. Instead always the context registry has to be used to lookup a particular context (even the parent).


Contextual Objects

The objects ("values") provided by the context are pure Java objects. They can be OSGi services, Eclipse extensions, Eclipse adapters from the IAdapterFactory or other contextual objects. Contextual objects are objects provided specifically for a context. A provider can be registered as an OSGi service which will provide context specific objects. To some extend, this can be compare to OSGi service factories. The only difference is, that the service implementation is not bound to the bundle requesting the service but to the context requesting the service. This allows for a concept of context singletons.


Context Registry & Security

The configuration of contexts is persisted and kept across sessions. A central registry is available of "loading" of contexts. However, access to the registry may be guarded by security constraints to allow only trusted code access to a specific set of contexts. This prevents client code with a lower lever of trust to not execute operations outside of the client context.

Back to the top