File system folder structure using type-namespace-id structure

Winery’s data is stored in a a file system ADR-0001. The contents of the repository should be

  • human readable

  • machine processable

Considered Alternatives

  • Folder structure using type-namespace-id

  • Everything in one folder. Hash-based storing similar to git.

Decision Outcome

Chosen Alternative: Folders subdivided in type-namespace-id

The final file system layout itself is documented at RepositoryLayout.

human readable

Everything in one directory causes many files listed and thus humans will have difficulties to find the right file.

The folders in the top level are the TOSCA “components”, e.g., Node Type, Relationship Type, Service Template, …

The second structuring element are namespaces. Namespaces are an established method to avoid naming conflicts and are a structuring element. TOSCA is an open system and everyone can create Node Types. One has no global control which names are given to Node Types. Thus, there might be two different Node Types with the same name. The namespaces provide a natural structuring and Winery reuses this idea.

The third structuring element are the ids of the respective definitions child (type, template, …): Each element has an id contained in the respective namespace. This id can be directly used as folder name.

Within each folder, the component-specific information is stored.

machine processable

Windows cannot create directories named Therefore, the names have to be encoded so that an appropriate folder can be generated.


Copyright (c) 2017 Contributors to the Eclipse Foundation

See the NOTICE file(s) distributed with this work for additional information regarding copyright ownership.

This program and the accompanying materials are made available under the terms of the Eclipse Public License 2.0 which is available at, or the Apache Software License 2.0 which is available at

SPDX-License-Identifier: EPL-2.0 OR Apache-2.0