Helping FOSS projects be more successful through clearly defined project data.

CC0

Becoming ClearlyDefined

Being ClearlyDefined can be more of an intent than an action. Many projects already make the necessary data (e.g., license, source location, attribution parties) available in easy to find places. Awesome! As long as the harvesting tools can find the info and you are consistent about keeping it uptodate, all is good.

For some however, there may be a few steps. One on of the core philosophies of ClearlyDefined is “meet them where they are”. That is, we are not looking to drive a particular behavior or format on your normal development process or repo structure. Rather we are aware of a host of conventions and we only need hints from you as to which approach you are taking.

Source location as a case-study

One key thing to note is that ClearlyDefined is about project releases, components. That is after all what people consume. Knowing the source location of the version of the component you are using is critical if you want to do targeted fixes, understand attributions, do deep code analysis such as security reviews, or even simply to fulfill the source disclosure obligations of copy-left licenses.

So whatever your release process, we are after two pieces of info: the repo location/type, and the exact revision that matches each released version. If you are using a structured packaging system like npm, this is pretty simple. There is a repository property in the package metadata and by convention the indicated Git repo will have a tag that matches the package version. Voila! No work. We just needed to know the convention and you just needed to follow it.

Our tools will do this kind of snooping automatically across a range of the data and formats supported.

Explicit data

In the npm case the convention at play is inferred. In other cases, approaches are ambiguous. For example, not everyone maintains their license information (e.g., the license itself or the list of attributable parties) in the same place or in the same way. Some projects want everyone in the gitlog attributed. For others it is all the copyright holders. For others it is the project itself. Dropping us a hint as to what approach you want clears up any ambiguity and makes it easy for the relevant data to be gathered.

You can leave that hint in a very simple clearly.yaml file in your release and/or in your repo. The details of this file are still being worked out. We actually aren’t intereested in foisting another file/format on the world so if your community already has a convention for where to store metadata, we just need to learn about that and then can look there for the hints.

Keep in mind that these hints are metadata – it would indicate how to find the license or attributable parties rather than give that information directly. That makes it an order of magnitude simpler and easier to manage. The information can go almost anywhere.

Help define this space

This is an evolving space and you can really help define how it works with the least possible impact on project teams.