A European Contributor’s First PGConf.dev
For years, the PostgreSQL developer conferences felt out of reach—located in Canada while I’m based in Europe. When the event moved to Vancouver for 2024, the distance only grew. But after hearing repeated encouragement at PGConf.EU in Riga, including a direct push from organizer Melanie Plageman, I finally committed to attending PGConf.dev 2026.
The journey itself was a time-zone illusion. Flying west from Frankfurt, we boarded at 10 am and landed at 10 am the same day, leaving a full Sunday to recover. A bike ride around Stanley Park on Monday offered the first chance to meet fellow attendees informally.
Community Sessions Kick Off the Week
Tuesday was dedicated to community discussions. The conference provided breakfast for all attendees, which naturally extended the hallway track—the most valuable part of any conference for exchanging ideas and connecting names to faces. Devrim and I ran a session on making PostgreSQL ecosystem packaging more visible. In hindsight, the scope was too narrow; a broader discussion on packaging and user needs would have drawn more productive input. A parallel session on packaging immutability challenges meant I couldn’t attend both.
Joe Conway later led a session on recognizing contributors. I demonstrated our “badges” project, which aims to acknowledge smaller contributions beyond the main contributor list on postgresql.org. Users would collect badges for achievements like “PostgreSQL 19 Contributor” or “PGConf.EU 2026 Speaker.” The patch for the django website is ready, and open technical questions were resolved during the session and subsequent hallway chats. Unfortunately, that slot overlapped with a TDE session I also wanted to attend.
The afternoon brought a session led by the PostgreSQL security team. They face a heavy stream of LLM-generated bug reports and are exploring better triage methods. One proposed outcome is shortening the current three-day release cadence to two days, or even one, to ship minor releases faster. Later, a panel of committers shared their most notable failures and unexpected successes—an engaging and humorous session. The evening ended with dinner in Gastown with fellow contributors.
Technical Talks and a Historic Round Table
Wednesday marked the first day of “real” conference talks. I attended sessions on index prefetching, scalability, and optimization. PostgreSQL 19 will feature what might be called “planner hints,” though Robert Haas presented an implementation that goes beyond simple hints. The day wrapped up with semi-joins and Ruohang Feng’s session on extensions—work closely related to Devrim’s and mine, suggesting future cooperation.
The conference highlight was a round table with original Postgres contributors: Bruce Momjian, Jan Wieck, Jolly Chen, Thomas Lockhart, Tom Lane, and Vadim Mikheev. They discussed the early days of POSTGRES in Stonebraker’s lab, how SQL was introduced, and the project’s management during the Postgres95 and PostgreSQL era. Melanie and Jonathan moderated expertly.
The social event took place at Vancouver’s Waterfront station, followed by a PG karaoke session that revealed some unexpected talents among attendees.
New Features, Deep Technical Sessions, and an Unconference
Thursday started with Joel Jacobson showing me the “key joins” SQL feature developed with Vik Fearing and others. What began as syntactic sugar for foreign key joins has become a theoretical framework on cardinality preservation in joins. A patch is underway where the optimizer proves query correctness, and the proposal has been submitted to the SQL standard committee.
Later sessions covered temporal queries with Paul Jungwirth, profiling PostgreSQL with perf (presented by Andres Freund), executor batching updates from Amit Langote, DDL deparsing with Masahiko Sawada, and pgstats in PostgreSQL 18 with Michael Paquier. The lightning talks were mixed—some excellent, others missing the mark with a room full of experts. The evening featured Paul Ramsey’s organized restaurant reservations, mixing attendees into new groups.
Friday centered on the unconference. Attendees proposed topics, voted on them, and the top twelve were scheduled. The most important session for me was led by Ants Aasma on consolidating TDE discussions. The community now has a clear plan for transparent data encryption, moving toward including it properly in PostgreSQL core. The conference closed with an extended tea break, keeping people around for final conversations.
Saturday I spent with family, hiking to Bowen lookout north of Vancouver. Sunday was a relaxed walk around the city before the flight back to Frankfurt.
PGConf.dev felt markedly different from the stressful PGConf.EU events I’d attended. Despite a fully packed schedule, the atmosphere was relaxing, with ample time for meaningful conversations. I typically jot down interesting topics at conferences and never revisit them. This time, I tried writing down fewer items—yet the list still grew to almost thirty. For once, I’m actually working through them. It was truly a worthwhile conference.
