BoF: Discovery and selection of code

[rachelss]

• How do we find software?
• How do we select software?
• How do you build software that is discoverable and reusable?
• How do we judge quality

[rachelss]

Existence

• We can search R repos to do something

Quality

• Benchmarked software suites - both existence and quality
• if in pipeline then that is a test
• does it have unit tests
• need a database of comparisons of output of software - ie benchmarking

[octopode]

Application-independent database structure is an initial requirement if multiple applications (i.e. assembly, alignment, orthogrouping, etc.) are eventually to be served by a central resource.

[DMaddison]

System like stack overflow / yelp for rating software that does particular jobs, with comments. Discoverability of software is hard - most people ask people in their "village" who to do; we often don't know what the people in the next village are using. Built it from the point of view of users. Contributions from users (which can be benchmark files plus pushing those through a suite of software packages and reports on results including various statistics) or software developers (e.g., they could run those benchmark files on their latest version). Could have a GitHub storage for the data (like Open Tree), but need a users' front end. People get credit for contributing to the system.