GSoC-19: Final summary

Developing a public data package recipe for retriever:

The Data Retriever was a package manager for data. Previously, the core software shipped with json script metadata. However, there was a need to shift it in a new location to help with maintenance and testing. retriever-recipes now holds all the scripts previously held by retriever, and also supports adding, editing and deleting of scripts.

A brief outline of the work done across the three phases of the project is mentioned below.

1. The first part of the project mainly comprised of shifting the scripts from the retriever repository to the new retriever-recipes repository. The other major task was moving the script adding and editing functionalities to the recipes repository. For doing so, necessary changes also needed to be done in the retriever repository.
2. Previously, Retriever would only look at the local scripts when needed to install or use a dataset. The goal of the second phase of the project was to enable downloading the scripts from upstream if not present locally. Later, I worked on the task of always attempting to use the latest version of scripts when installing or using a dataset.
3. In the last phase of the project, I made the necessary changes to the two Retriever wrapper repositories: Retriever.jl and rdataretriever. Apart from it, I opened pull requests for getting things merged into the master branch from the current development. I also performed exhaustive checking for errors in documentation and code.

During the course of the project, I mainly contributed to two repositories: retriever and retriever-recipes. The list of commits made in the two repositories:

• Commits in retriever
• Commits in retriever-recipes

The list of major pull requests made for the project:

• GSoC: Add provenance and recipes
• Merge retriever-recipes-dev into master
• Make appropriate changes for recipes features
• Make changes corresponding to retriever

I also wrote a number of blogs writing about the work done, on a bi-weekly basis. The list of the blogs written for the GSoC project so far:

• GSoC-19 with Weecology
• Retriever Scripts repository
• Retriever changes
• Download scripts only when specifically needed
• Install datasets using the latest version of scripts
• Julia and R repositories

In the end, I would like to thank my mentors Ethan White and Andrew Zhang for giving me this opportunity to work with them. Also, a special thanks to Henry Senyondo for his constant support and guidance. You were always there to help me when I was stuck somewhere and helped a lot in making crucial design decisions for the project. Would surely continue contributing to the project in future. Thanks to all.