class: center, middle, inverse # DIRAC and open source, take N ## Radovan Bast [@\_\_radovan](https://twitter.com/__radovan) Nordic e-Infrastructure Collaboration/ UiT The Arctic University of Norway --- class: center, middle, inverse ## Disclaimer ## - Current state has a history: I am not criticizing past, I want to influence present/future ## - My career will not depend on this discussion and our decisions today --- ## What I want to achieve <img src="https://cdn.jsdelivr.net/gh/bast/dirac-meetings@a64a67b8989ff5ce08dd416df59fe271556c1be4/img/forking-workflow.jpg" style="width: 100%;"/> - I would like the central master is public and licensed under LGPL - No restrictions for your developments, you can have private feature branches --- ## Paper freedom vs. software freedom ### Sharing papers - We want maximum visibility and maximum reuse. - The more interesting science is done referencing my paper, the better for me. ### Software freedom Is the freedom to ... - ... run the software for .emph[any purpose] - ... .emph[study] how the software works and to adapt it to your needs - ... .emph[redistribute] copies of the software - ... .emph[improve] the software and distribute your improvements to the public ### My view: giving others this freedom is good for us --- ## Typical concerns .quote["I cannot afford to share my own code"] - You don't have to, keep it on a private branch. .quote["I worry that others will copy-paste our code into their programs"] - Does our current license protect against this? Does our current license encourage/require sharing improvements done using our code? No. .quote["But we already share the sources - what else do you want?"] - Making sources readable does not mean it's open source. Our license discourages derivative work and probably limits our impact and funding opportunities. --- class: center, middle, inverse ## **If** you distribute sources, **never** design your own license, **unless** you have expertise in copyright law. --- ## [Unpopular opinion:] Our license is incompatible with open science This licence for the Dirac program suite (home page http://diracprogram.org) is valid for all use of the Dirac releases from 12. December 2018. The Dirac program suite is a free relativistic computational quantum chemistry program which is available as source code for all who accepts our license conditions: 1. All research publications including results obtained with Dirac must reference the Dirac program as specified on http://diracprogram.org. 2. No part of the Dirac code must be included in other software .emph[without written permission] from the Dirac authors. 3. The Dirac .emph[source code may be modified] for the user's research purposes under the conditions that: - it is clearly stated in all publications using the modified features that it is from a locally modified version of Dirac - the modified Dirac source code .emph[must not be further distributed without permission] from the Dirac authors ### Read this through the eyes of an external research group or funding agency --- ## You have to worry about software licensing <img src="https://cdn.jsdelivr.net/gh/bast/dirac-meetings@a64a67b8989ff5ce08dd416df59fe271556c1be4/img/in-out.jpg" style="width: 100%;"/> .cite[Idea: R. Darst, adapted] - In few years "available upon request" and "we have used a locally modified version of DIRAC" will not be publishable. - Sometimes "OTHERS" are you yourself in the future. --- ## Concrete steps towards public master under LGPL - Get agreement from all present and past contributors - Move `MOD_UNRELEASED` code out of master - Possibly rewrite Git history to get rid of these changes in the past ### Two (or more) repositories (like Dalton) - https://gitlab.com/dirac/dirac/ (only master and release branches) - https://gitlab.com/dirac/dirac-private/ (all other branches, no master branch) - https://gitlab.com/john-doe/dirac/ (no problem to fork and work wherever)