Just kidding! 
But it can’t hurt to dream. In the meantime, we can appreciate the nice tradition. Over on the Ubuntu shores, we’ve geared up for another year of getting docs in shape. We have lots of plans for improving and extending existing docs, as well as for publishing completely new content and continuing to tighten the Ubuntu docs story overall.
Keeping up with the times
@sally-makin has been running experiments with GitHub Copilot, using the Ubuntu Server docs as her playground. Her goal? To understand the differences between the different LLMs, what they can and can’t do well, and how they could be harnessed to better serve our documentation and our readers. Like any good scientist, she’s recorded her results for your enjoyment. See it at Ubuntu Server Gazette – Issue 11 – AI-bots shall eat my docs…
Speaking of robots…
I’ve been looking into ways to make our docs better accessible (consumable, parseable) by LLMs. A number of people suggested using llms.txt, which is a proposal “to standardise on using an /llms.txt file to provide information to help LLMs use a website at inference time”.
The idea is to make it simpler for LLMs to parse website content (in raw Markdown as opposed to HTML – which is a pain to parse anyway – with all the stuff that’s not really relevant, such as navigation elements, headers, footers, etc.).
As long as we accept that models will consume our docs and use them to appear smart, we might as well do our best to present the content in a way that makes the consuming easy and efficient (and in so doing hope that the regurgitated AI output based on our docs will be of high(er) quality).
A couple of extensions exist that seamlessly enable support for llms.txt for Sphinx documentation sites. I looked into sphinx-llm and sphinx-llms-txt. Their output is pretty much identical (which is good), but they both have some drawbacks. I also ran a bunch of benchmarks to evaluate how impactful this would be (i.e. how much time/tokens it would spare compared to an LLM parsing an HTML site).
I expect to write a separate post about this investigation where I’ll detail the pros and cons of the two extensions, as well the measured results from the benchmarks.
Ubuntu Server docs push towards 26.04 LTS
The Server team recently held a documentation sprint to get the docs ready for the 26.04 LTS. Sally set out an ambitious goal of triaging, checking, and testing every page in the documentation (or as many as they could) to make sure they’re correct and up to date for 26.04. As a consequence of this event, they also filled up their backlog for things that are not urgent to fix before 26.04 — which gives them plenty of documentation improvements to make for 26.10.
Sally was also using it as an excuse to do more Copilot tests on her friends and colleagues (agent reviews of their pull requests). She’ll be publishing more details of those tests next time!
Ubuntu wiki: the king is (almost) dead; long live the king
After the initial announcement of the Ubuntu wiki project, @shanecrowley made a second post about the features to expect in the new wiki, and has been responding to feedback from the Ubuntu community. It’s been great to see that people are excited about what we’re working on.
Enterprise Ubuntu plug
Since the launch of Ubuntu Pro for WSL, Shane has updated the deployment tutorial to provide users basic guidance on using Landscape SaaS and also updated the Pro for WSL installation guide with the multiple installation options that are now available.
The docs homepages for both WSL and authd have also been restructured into the new homepage pattern being adopted across the Canonical documentation, which better guides users to specific domains-of-concern.
What’s coming up?
This spring is all about getting 26.04 LTS (Resolute Raccoon) ready. The Feature Freeze looms nigh (Feb 19), and right after that, the Foundations team will be holding their Docs Focus Week. The idea is to close as many bugs as possible from the Ubuntu Project docs backlog, as well as to publish improvements and additions to the Ubuntu for developers docs.
For those who will be upgrading their Ubuntu installation, I also have a merge proposal pending to improve the guidance provided by the ubuntu-release-upgrader. Fingers crossed it makes it in for the release.