Open Source Working Group RIPE 87
30 Nov 2023, 09:00 (UTC+1)
WG co-Chairs: Martin Winter and Marcos Sanz
Scribe: Razvan Oprea
Status: Final
Martin and Marcos welcomed the audience and opened the meeting. No new items were added to the agenda, and the minutes from the previous meeting were approved.
The co-chairs thanked the community for their interest in the co-chair position, having three candidates for the seat available.
The three candidates, Christian Scheele, Luka Perkov, and Sasha Romijn, came on stage in turns to introduce themselves.
Maria Matejka suggested that the new WG co-chair should aim to move the Open Source WG to the afternoon.
Peter Ballot, WG Chair IoT, speaking for himself, expressed his support for Christian Scheele.
Valerie Aurora, Frame Shift Consulting
The presentation is available at:
https://ripe87.ripe.net/wp-content/uploads/presentations/90-2023-11-30-Credit-for-Open-Source-Software-Contributions-6.pdf
Tim Wattenberg, DECIX, agreed that first-time contributors should get credit. He also questioned whether people give credit to the open-source projects they are using.
Alexander Zubkov, QRATOR LABS, commented via Meetecho that policies should be short; otherwise, they could stop contributions.
Gert Döring, SpaceNet AG, agreed that policies should be short. He added that assigning credit for multiple patches can be complicated and described how this is done in the OpenVPN project.
Nat Morris, NetBox Labs Inc., asked what Valerie thinks of projects having an author’s file. Valerie responded that any method for giving credit makes sense.
Nat also asked whether licensing agreements are off‑putting for first‑time contributors. Valerie agreed that they were.
Maria Matejka, CZ.NIC questioned whether a one-line code is plagiarism. She argued that a one-byte commit may result from five weeks of work. Valerie agreed.
Martin Winter stated that beyond the people who contribute code, there are people who review and comment on code, and he is also looking for ideas on getting them recognised. Valerie said that code review is in short supply because it is not rewarded. Gert Döring then explained how code reviewers get credit in the OpenVPN project.
The presentation is available at:
https://ripe87.ripe.net/wp-content/uploads/presentations/62-2023_11_30_ripe_oss_external_submissions.pdf
Maria ended her presentation with three questions for discussion:
“Do you have programmer’s doc? … is it up-to-date?”
“Do you like incoming patches? … and merge them?”
“What other issues are you dealing with?”
Valerie Aurora, Frame Shift Consulting: commented on how the Linux kernel project deals with code merges and suggested that code reviews should be done as a group.
Maria responded by acknowledging that reviews are hard and explaining how a policy that states conditions for merging might make the process easier.
Marcos Sanz agreed, arguing that besides pointing at the documentation, a policy should prevent much time spent repeatedly explaining the same things.
Maria agreed, stating that it all comes back to having programmer’s documentation.
Martin Winter commented that testing and documentation should accompany patches introducing new features, which might put some people off. He also wondered if any project had the programmer’s documentation up to date.
Maria questioned whether full credit should be given to authors who do not submit documentation or commit messages that must then be written by someone else.
Sasha Romijn, Reliably Coded, admitted that IRRd documentation is no longer accurate. They added that one of the challenges is accepting new code that introduces a feature which must be maintained for years because paying customers may come to depend on it.
Maria agreed that it was a big problem.
Martin Winter, NetDEF / OpenSourceRouting
The presentation is available at:
https://ripe87.ripe.net/wp-content/uploads/presentations/96-testing_open_source.pdf
Gert Döring, SpaceNet AG, commented that the OpenVPN project does not have so many resources for testing. He then referred to the environmental impact of the test environment, suggesting the CI should evolve, for instance, by running a subset of tests upon initial patch submission and maybe running the full set of tests once a day.
Martin replied that power consumption was a very interesting topic and said that maybe the first step is figuring out the energy consumption of a single test run. As for deciding what tests should run, he agreed it was a good idea but said he did not know how to do that reliably and automatically and was open to ideas.
Alexander Zubkov, QRATOR LABS, wondered whether the infrastructure or the people cost more.
Martin answered that the people cost more, the hardware being far cheaper.
Μartin announced that the voting on the mailing list was open and stated that they would do their best to move the session to the afternoon even before the new co-chair joined.
The meeting was closed.