![](\"\/images\/journalpics\/062026\/oss_2026_quantum_keynote_sm.jpg\")
<\/a><\/div>\nThe next few talks I attended were around Open Source Program Offices (OSPOs), the first was a panel around scaling them, where I was happy to see a wide range of experiences presented from across industries, including Ford, Geico, and CISCO, with a moderator from GitHub. There’s definitely a trend of using AI tooling to help the work in OSPOs, especially as the work and importance of these organizations grow. I also enjoyed Dawn Foster’s talk on “Strategic Approach To Demonstrating the Value of OSS Efforts” where she presented the highlights of the Demonstrating Organizational Value<\/a> Practitioner’s Guide that she co-wrote.<\/p>\n From there I dipped into a couple of MCP talks, starting with Automating MCP Server Testing<\/em> from Neethu Elizabeth Simon of Arm. It was an interesting talk overall, but my big takeaway was the server she demonstrated with: the Arm MCP Server<\/a>, which is “An MCP server providing AI assistants with tools and knowledge for Arm architecture development, migration, and optimization.” Cool. I took notes and was texting with Rishi about how we might be able to leverage something similar in the s390x world. From there I went to a talk on ContextForge<\/a> in the context of managing MCP sprawl. It was also a timely talk for me since we had just on-boarded the project to our GitHub Actions runner a few weeks before and I’d been meaning to learn more about it. Done! It was also nice seeing Dave Neary and his What Developers Should Know About Hardware Architecture<\/em> talk, where he throws out lots of interesting facts about things that matter in architectures, along with other topics like just how slow spinning disks are, especially compared to CPU caches.<\/p>\n Monday evening was the conference reception at the Mill City Museum<\/a>. It’s a museum I had wanted to go to anyway, so I was happy to learn that was the reception venue. It was a nice museum about the history of flour milling in the city, and the elevator they converted into a “ride” that takes you up and down several floors through a history of the mill was an absolute delight. The views from the top where you’re let out were beautiful, especially since the rain hadn’t rolled in yet! Unfortunately, the rain did eventually come in, just as the scheduled drone show for the 35th anniversary of Linux was to start. The drone show still happened, but it wasn’t really dark enough and drizzle quickly turned into a thunderstorm.<\/p>\n Tuesday’s keynotes got us back in the realm of AI, with talks on Strands Agents from AWS and AI used in robotics. They also had a panel on the value of open source software foundations for projects, which surprised me as a keynote topic for an open source conference, because most of what was covered struck me as pretty basic, and I think a lot of folks in the audience had experience engaging with projects within foundations.<\/p>\n From there I went to a talk on Package Testing Across Distributions and Architectures at Scale<\/em> where the presenter walked through a move away from dozens of servers and consolidated on x86 and arm64-based cloud servers, using Ansible Molecule. Alas, not immediately applicable in our s390x porting work, but it was an interesting tour of what they had done and how much disk speed impacts package building performance. An afternoon session from the OpenSSF that gave a whirlwind tour of their projects was also interesting, since it can be a little tricky to keep up if your day to day work is unrelated to security.<\/p>\n That evening was the Tux Trek in the expo hall, and in addition to some great conversations, including one where I got to nerd out about mainframe stuff to someone who was curious, I got a tux cotton candy! I then got to have dinner with Rishi and a colleague who I didn’t expect to see there: Cheryl Fillekes! She’s worked on various OpenShift products for Linux on IBM Z, and we’ve connected on various topics over the past several months (mostly when I needed something, hah!). It was really nice to meet her in person. I was also pleasantly surprised when she suggested our dinner adventure should include finding some Guinness 0 (non-alcoholic). You see, I’m a beer fan, but I got a bit sick back in March and have been taking a break from drinking until I feel better. Guinness 0 was a nice treat, and surprisingly good!<\/p>\n Day three began with the obligatory keynote discussion with Linus Torvalds, followed by another keynote about AI agents, but this time with a focus on specifically how they can help open source software maintainers. I really enjoyed Kate Stewart’s keynote on the Zephyr project though. Zephyr has been on my radar for a while, but playing with it is another one of those projects I am not sure I’ll get around to anytime soon. But hey, it turns out my RISC-V board is supported by Zephyr! That gets me one step closer!<\/p>\n The final keynote on Free to Use, Not Free to Run: Reinventing Package Registries<\/em> is one that stuck with me. Basically, open source communities are running registries (PyPI, RubyGems, crates.io, etc) that companies using the ecosystem depend upon, but they’re becoming expensive to run, and projects are now hyper-aware of security and supply chain concerns that they need to take very seriously. So they’ve launched a Sustaining Package Registries Working Group to discuss concerns, and develop solutions together, up to and including starting to charge for industrial scale traffic. I’ve seen the thankless work that goes into things like PyPI. Helpfully, Stephen wrote about this too: 10 trillion downloads are crushing open-source repositories – here’s what they’re doing about it<\/a>.<\/p>\n Most of the rest of my day was spent meeting up with people, but I also happy to make it to Rob Landley’s Building the Simplest Possible Linux System<\/em> talk. I’ve seen Rob speak a few times over the years, and I love the enthusiasm for the work that he brings to it. He’s also been friendly in the realm of s390x porting, and I was delighted to see on his screen as he was preparing to begin that the emulated build he had up on the screen was s390x.<\/p>\n![](\"\/images\/journalpics\/062026\/oss_2026_ospos_panel_sm.jpg\")
<\/a><\/div>\n![](\"\/images\/journalpics\/062026\/oss_2026_arm_mcp_sm.jpg\")
<\/a><\/div>\n![](\"\/images\/journalpics\/062026\/oss_2026_reception_roof_view_sm.jpg\")
<\/a><\/div>\n![](\"\/images\/journalpics\/062026\/oss_2026_zephyr_keynote_sm.jpg\")
<\/a><\/div>\n
![](\"\/images\/journalpics\/062026\/oss_2026_small_linux_talk_sm.jpg\")
<\/a><\/div>\n![](\"\/images\/journalpics\/062026\/oss_2026_selfies_sm.jpg\")
<\/a><\/div>\n