Gunnar Wolf
salsa profile: https://salsa.debian.org/gwolf/
gitlab profile: https://gitlab.com/gunnarwolf/
github profile: https://github.com/gwolf
Debian Developer since 2003, and (still relatively early) in the process of joining the Debian Greybeards team. Have been involved in various bits of the project; currently my main focus is in ① keyring maintenance, ② DebConf organization, ③ mentoring a GSoC contributor to adopt and update the Raspberry Pi images creation.
Accepted Talks:
DebConf Committee BoF
Meet the DebConf committee. See and hear us discussing this years and next years’ DebConf. Give us feedback and ask for improvements.
DebConf 26 BoF
We have the fine tradition of having DebConf N share their experiences with DebConf N+1. The DebConf committee will also be present and we invite a core member of every team to come around and help with the knowledge transfer.
The session will begin with a DebConf26 presentation.
DebConf 27: In your city?
Session to discuss possible locations for DebConf27!
DebConf 26 is planned to be held in Santa Fe, Argentina. But we have no plans for DebConf27, yet.
Continuous Key-Signing Party introduction
One of DebConf’s recurring activities is the Key Signing Party. It helps Debian strengthen and expand its web of trust. This session will:
- Explain how keysigning is done in a DebConf setting
- Validate the SHA256 hash of the KSP coordination document
- Explain how to participate to people who did not send their keys in time
Once DebConf is closer, we will link from this talk proposal to the relevant documents you should have in hand for the keysigning party.
A retrieval-augmented-generation pipeline to help users query system-provided documentation
The increasing integration of AI into computing workflows demands a re-evaluation of traditional operating system design. In environments like Debian, users are often faced with a vast ecosystem of command-line tools, each accompanied by extensive manual pages (man pages) detailing usage, flags, and parameters. While comprehensive, these documents are frequently dense, verbose, and not well-suited for rapid onboarding or targeted queries. We propose a retrieval-augmented generation (RAG) pipeline to bridge this gap, enabling natural language interaction with system documentation. By combining tokenization, embedding, and dense retrieval with a language generation model, our system allows users to query tool usage in plain language and receive concise, contextually relevant responses. This approach streamlines tool discovery and comprehension, and represents a step toward more intelligent, user-aware operating systems.
Closing Ceremony
We say farewell to DebConf 25 and look forward to DebConf 26!