Public:BWmeta service/Transformers

From YaddaWiki

Revision as of 08:07, 19 November 2010

Background and motivation

As of 2010Q4, three versions of BWmeta format and at least three Java models of BWmeta are used in YADDA. See here for details. Serialization and deserialization is implemented for the following pairs:

	Repo	DL	Y	…
1.0.5
1.2.0
2.0.0

There are also converters between selected models:

• Y ↔ DL

Serialization of Y model to BWmeta 1.0.5 is therefore possible through the following chain of transformations:

• Y model → DL model → BWmeta 1.0.5

Conversion from DL model to Repo model is possible through:

• DL model → BWmeta 1.0.5 → Repo model

Unfortunately, many mutually-incompatible transformation interfaces are present. It is difficult to understand which transformations are possible, how to combine them, which classes should be used for that end and how to instantiate the classes.

A unified interface to all the transformations would be very helpful.

Unified interface (under construction)

Three different transformer types are present: readers, writers and converters. Each type has a corresponding interface:

• IMetadataReader<T> deserializes an object of type T from a String or a Reader
• IMetadataWriter<T> serializes an object of type T to a String or a Writer
• IMetadataConverter<S, T> converts an object of type S to an object of type T

MetadataTransformerFactory maintains a list of available "atomic" transformers and is able to combine the atomic transformers into more complex ones.