Reuse Rating F.A.Q

There can be several reasons for this. Most often it’s because

> your project is an end-user application which is not likely to be embedded, and is not explicitly reused as a result, or

> the uploaded file is not characteristic of your project or that release.

We also observed that projects which tend to release several patched versions will see low reuse rating at the release level – the reuse will be diluted over several releases. In this case, displaying all measures in point form will fix this problem. Likewise, integrating a new release into a project can take several weeks, or may not happen at all.

The measure of reputation, based on monitoring the rate of reuse within a large set of open source projects, is a byproduct of our R&D program. As a frequent user of open source libraries and applications, and in the belief that it could be useful to the community, we thought it natural to provide this service for free.

Antelink is a software vendor. Our market position leaves little room for developing service/consulting activities. As a result, we didn’t opt for an open source model. As users of open source libraries and solutions, we felt it was important to give something back to open source communities. Part of this will involve developing products  and services that can be used for free by open source communities.

Many classifications rank open source projects according to the number of developers, forum activity, or the number of downloads. One metric alone is always questionable and subject to interpretation. Combining several, however, provides a ‘big picture’. New metrics are useful in that they make it possible to characterize new uses.

We take a file and measure the amount of reuse in a large database of open source projects. Reuse can be direct, involving the explicit integration of the file, or indirect via calls. We improve our algorithms based on the scenarios we encounter, so feel free to share your experiences sending us feed-back.

We do measure reuse, and naturally focus on libraries and forked projects. Open source projects dedicated to end-user type applications will not be highly rated.

Initially, we wanted to start with a representative sample. In terms of the number of hosted projects, the two main forges are Sourceforge and GoogleCode. We are currently downloading content from 162,608 projects – an interesting start in our view. In practice, we use other sources as well such as savannah, RubyForge and OW2.

Our engine continually picks up new contents and updates data on projects which are already catalogued. The frequency of updates varies between a few days and a few weeks based on empirical rules.

We focus part of our R&D work on ensuring the robustness of our metric. This involves taking into account uses which may distort results by counting the same reuse several times. For example, we automatically identify duplicate projects found on both SourceForce and GoogleCode. They are only counted once.

In principle, any type of binary file (jar, tar, png, jpg, etc) or text (code source, readme, licence, etc.). The measure is not used in exactly the same way for every case of course. Special attention must be paid to content associated with reuse (.jar, .dll, .h, .css, .png, etc). The only restriction is on file size (100B – 100Mb max.).