OggCamp26, Manchester
No one knows what an unconference is before they go to an unconference. Even as I'm writing this the spell-checker is confused: "Do you mean conference?"
The definition of an unconference is that you show up, other people show up too, and you all talk about what you each care about.
"What if everyone there has nothing in common with you?", quipped my partner as I explained my intention to buy tickets to an unconference.
To avoid this worry, OggCamp26 had scheduled talks. At the very minimum, you'd go to a talk you like and get some value from that.
I wanted to go to these talks:
- Exploring Algorithmic Art (using Turtlestitch), Margaret Low
- Infrastructure as Code is Cool, Charlie O’Hara
- Making Minecraft maps with science!, Michael Dales
Algorithmic Art
I made it to Margaret's talk and discovered that children learn maths and algorithms via stitching. The physicality of the process / output creates a better memory of the algorithm used versus code on a screen. I also heard that 3D printers in schools are more hassle than they're worth, and that a modern sewing/embroidery machine can be easier to use, maintain, and can drive better educational outcomes.
I always forget visual patterns I've previously created and will use my local hackspace embroidery machine to validate the recall idea. Planning to test this with Anki for a full statistical analysis.
My homework is to review art from Frieder Nake, Paul Klee, Clive Richards, Waclaw Szpakowski, visit the V&A Museum digitally and in person, and review mathematical patterns like the tatami stitch, diamond pattern, tilling patterns and more.
Observability
Everyone else made me jealous that I didn't make it to the rest of the talks, and I'm mostly kicking myself about missing Charlie's talk: an Ansible showcase and live demo spinning up, tearing down, and redeploying the unconference website. Using Ansible at home for years has been great, but I still think I need to do more with it, and this talk is full of tips, tricks and laughs.
Circling back to the unconference thing. What did I want to bring to it? Of course some might argue that volunteering was enough, but what else.
After helping the crew with setup on Friday I decided I'd give two talks and try to exercise my public speaking. My recent obsession with documenting projects and packaging them into nice repos paid off, because I could easily whip up a talk from my blog posts.
I spoke about homelab container observability with the Grafana stack and highlighted Alloy, the newish runner from Grafana that's taking center stage since the deprecation of Promtail.
Prometheus and Mimir were used in the stack and this caused some confusion: why have both in a stack?. The idea with the stack was to showcase best practice, and Prometheus best practice is short term log storage. You can fully migrate to Mimir for both short term and long term storage needs, but here I wanted to highlight best use for Prometheus, with the upgrade path being Mimir. You don't need both, but you shouldn't have long term storage with Prometheus alone.
After the talk I found out that people were using the Grafana stack to cutback enterprise DataDog costs and that Alloy is enabling clever stacks for feature parity at a fraction of the cost. I'd been working on this observability repo in isolation and the fact that people were using it helped me justify the time spent.
Sports tracking
My second talk was on how to use Strava data with GeoPandas and OpenStreetMap, for visualisation, analysis, and route suggestions.
Most people expected me to show them how to build a FOSS Strava, and maybe I will someday. I showed an old project that detailed how to: connect to Strava via OAuth, download your own data, check the coordinate system of that data, convert it to OpenStreetMap coordinates, use OSMnx library to display the routes on OpenStreetMap, and finally use the OpenStreetMap road network to suggest alternative paths along the route. For FOSS data collection, I highlighted OpenTracksApp instead, a good offline activity tracking app from F-Droid.
Some attendees couldn't find the OpenTracksApp on the Play store and we started showcasing F-Droid, LineageOS, GrapheneOS, and the best cheap hardware to use it them on. We later spotted an F-Droid core contributor, snapped their fractar t-shirt design (one of many collectible patterns), and bargained for a bunch of stickers.
Although I've used OpenTracksApp exclusively of late, I've not done any statistical comparisons vs Strava or Garmin. I'd like to answer what is the accuracy variance between the three and is it acceptable enough to justify the exclusive use of OpenTracksApp.
After the session some folks showed me their OpenTracksApp data, hundreds of cycling miles. We quickly realised the need to get this data out of OpenTracksApp and into a database for local backups and future analysis. It made me more comfortable with the accuracy of OpenTracksApp.
The repo contains a GeoPandas example for visualization and I explained that you can add Pandas and leverage it for queries and charts. People also wanted to use the Strava, Garmin, or OpenTracksApp data to ask complex questions via advanced sports statistical tools like Golden Cheetah.
Fun conversations
Running along the observability track I spoke to people about NixOS, how it can become a configuration hellscape, and how it might be useful for making custom system ISOs that reduce the need for Ansible playbooks that set everything up. Some used NixOS, some dreaded the idea of it.
Kyle and Jasper showed me GNU Stow, and how people use it to store their system configs. I'll have to give it a try.
Lee, gave a talk about his UNIX book "The Ghost of SVR4" and most folks I spoke to used Linux as a daily driver or at minimum tried it previously. Linux Mint was a favorite amongst attendees.
Pete told stories about HAM radio, using radios whilst hiking for fun and emergency scenarios, and how MeshCore is better than Meshtastic. He rocked his own MeshCore node, but he wasn't alone. A handful of other nodes cropped up during the event and we got to speak to them too. There was even an unconference talk on radios too. As a result, I've already put a MeshCore device on my Christmas present list.
Trig points (Triangulation station, Wikipedia) and trig-bagging (https://trigbagging.co.uk/), came up too.
The team at Kraken, fed us lunch and showed us their offices, and talked about the Open Charge Protocol (OCPP) , and ROSCO toolset.
The unconference talk on retro tech featured Psion devices, and it emboldened my craving for a Psion Series 5 which can run OpenPsion, Linux for Psion.
In general people spoke about old Blackberry's, new Android phones with Blackberry keyboards, IBM and Lenovo Thinkpads, refurbished Microsoft Surfaces, and used Android phones.
Tim talked about using DNS lookups for data exfiltration, and dns.toys.
Most people were self-hosting, using NextCloud, using Proxmox, using Immich, and especially Home Assistant. The home assistant talks were oversubscribed, incredibly interesting, and core contributors where there too. I'm now on the lookout for a home assistant compatible extension lead, with built in power meters.
A Yocto Project contributor was there too and we spoke of hard embedded hardware / software bugs, AWS Greengrass, and the Yocto Project itself.
On data science: the general consensus was that R is still an interesting language to learn and use for data science, whilst Python is more of a general purpose language.
On databases: there was a bit of chat of DuckDB and Postgress, with me plugging CouchDB+PouchDB after a recent pleasant experience with it.
Wish I would've seen:
- Shane's talk on using Godot for game development - Shane from Linux Lads Podcast fame
- Jasper's talk on "Usability in Open Software"
- Millie's talk on "Home (security) Assistant"
- Talk on XMPP
To Unconference, or ...
In the end, OggCamp26 was about belonging, and during the event I met and talked to people that cared about a lot of things that I care about.
The intersection of other attendee interests and my own was very large, to point where I'd have difficulty picking enough single topic conferences or events that can match.
To get the same results I'd have to attend:
- an observability or devops conference
- a Rust conference
- a data science conference
- a geospatial conference
- an embedded conference
- a radio and LoRa conference
- a retro-computing conference
- an IoT conference
- a Linux conference
- an Android conference
- a digital art conference
- a gaming conference
OggCamp26 was ace because it opened conversations across all of those topics. If OggCamp26 were an all you can eat buffet it would be one that carries a lot of dishes from around the world with each being the best version of that dish: the best pizza, the best curry, the best ramen, and the best schnitzel.
Who wouldn't like that? And in terms of value: n+1 topics for the price of one conference ticket?
Sign me up!
You can read more about OggCamp26 and see the schedule here, OggCamp26, Schedule. The main track talks are all online.
And I hope to see you at OggCamp27.
Other opinions
Other attendees have shared their thoughts about the event: