rcomp is a proof-of-concept for facilitating reproducible research by running primary assets remotely while providing interaction and result collection by easy-to-install client software.
Source code for the server is in the directory server/. Note that it is
collectively referred to as the server, but in general it can be deployed
over multiple hosts, e.g., with reverse proxies that load balance across hosts
that have rcompserv processes.
Source code for clients are in the directory frontend/. They are
approximately sorted according to programming language, e.g., the directory
py/ contains a client library in Python, and the directory js/ contains
a client library in JavaScript (ECMAScript) with examples using Node.js.
Note that some clients have Web UIs, so their installation and execution involve
"backend" components, e.g., nginx configurations.
Version numbers are managed independently among clients and the server. One motivation to do this is some parts will be available as separate packages depending on best practices for respective platforms, e.g., releases of the Python client will be available on PyPI. Allowing version numbers to diverge enables us to enforce that changes to the version number of a released package only occur when code in that package changes. However, tracking large gaps among component version numbers or attempting to maintain backward compatibility far into the past does not yet have practical motivation, so major and minor version numbers will more or less increase at the same pace.
This is free software released under the terms of the BSD 3-Clause License. There is no warranty; not even for merchantability or fitness for a particular purpose. Consult LICENSE.txt for copying conditions.