Mark Moxon has decompiled Elite, and has blueprints for each of the ships, which I used (with his permission) for the build

I wanted to recreate the wire-frame polygon aesthetic of the original. I wasn’t that sure how best to achieve this; eventually I settled on 3d printed brackets for each vertex, and joining these with rods. At this stage, I wasn’t really sure where I was going next…

There was a scheduled track, as well as two unconference tracks. Some of the talks for these went up very late, and I missed (most of) a couple I’d have liked to have seen.

Some highlights, in no particular order:

I caught the tail end of William Morland’s talk on Home Assistant, and discussion of wiring up boilers to it. Apparently Dutch gas boilers have to have an Operntherm interface, so something to consider when mine needs replacing.

There were only a couple of lightning talks unfortunately. There was an interesting one about an LED score-board for unicycle hockey(!), and using DNS exfiltration to get save the scores (and get the time) over a locked down wireless network.

Next up was a great talk on keeping an 80s laser tag system alive

Margaret Low gave a good talk on Turtlestitch (essentially Logo for embroidery). She spoke on this at the last Oggcamp, which had inspired me to go down a rabbit hole of making this, so it was nice to be able to say thanks for the inspiration.

There was an incredibly depressing talk about whistleblowing and content moderation in Big Tech; really bleak.

The retro computing show and tell was good - mostly Psions and Z88s

There was a great talk on repairing Dyson battery packs (which involved writing custom firmware to unbrick them).

One of the highlights was Charlie O’Hara’s on infrastructure as code. It was a great demo of Ansible, and concluded with a spontaneous (and somewhat nerve-wracking) “shall I delete my website live on stage and restore it”? (answers: yes and yes)

The whole event was great; really friendly and interesting people. (Maybe ~150 in total?)

This year the main track was recorded.

A great warm up for EMFCamp :-)

I’m hoping to be able to write my own designs to a card, but haven’t got that far yet. As it’s rather a long write up, and still a work in progress I’ve put the write up on a separate page.

I’m still not quite sure what went wrong. It think I’d somehow ended up with (only) the 22.04 kernel installed, which led to lots of “Unknown symbol” errors. My system was quite heavily customised, so it was perhaps a bit optimistic to expect the upgrade to work.

This post is mostly a note to self about how I reconfigured the dual boot setup in case I need to do it again (which I hope I don’t).

I initially thought it was only the kernel that was the problem. This stackexchange post was really useful for mounting the encrypted drive, setting up a chroot environment and reinstalling the kernel. That got me booting to the command prompt.

It turned out I was missing most packages, at which point I decided to wipe and restore from backups.

(I could actually access my encrypted drive through the Ubuntu LiveCDUSB, so took a fresh backup of /home and /etc from this)

I’d got an (essentially unused and tiny) Windows 10 partition dual booting with Ubuntu. I was keen to keep Windows 10 on the machine, as it’s occasionally useful. I’d previously set this up to dual boot with full disk encryption for both OSs roughly following the guide here, but using the legacy BIOS, as that’s what Windows 10 had come preinstalled with.

I was pleased to see that the guide had been updated for Ubuntu 24.04 and is now much easier. I took this opportunity to wipe the Windows partition and switch to UEFI too.

The basic process was:

• Use Ubuntu 24.04 USB image to check backups are OK and work
• Use the Ubuntu USB image and gparted to remove all partitions
• Reboot and set the BIOS to UEFI mode (only) (Startup menu in BIOS), and check the security chip is set to “Intel PTT”.
• Install Windows 10 Pro from installation USB (avoiding creating a Microsoft account by selecting “Work or school account” then “Domain join instead” via
• Install interminable updates and reboot lots of times (I kept force checking for updates until there were no more). (Bitlocker is not enabled at this point)
• Reboot with the Ubuntu USB image
• Follow the instructions at https://www.mikekasberg.com/blog/2024/05/20/dual-boot-ubuntu-24-04-and-windows-with-encryption.html to setup the dual boot with an encrypted partition setup.
• Check both still boot
• Enable bitlocker on the Windows partition You need to boot Windows via the BIOS’s boot menu (F12 on my machine), not via Grub, otherwise it will prompt you for the bitlocker recovery key on boot (details of how I removed the Windows option from Grub below)

Once everything was working, I rebooted using the USB image to mount my nfs share containing the backups and restored my data.

A few post install tweaks:

Hibernate doesn’t work out of the box. I enabled this following the guide on Elbrar’s corner. This requires booting with the USB image again, to resize volumes to make space for the encrypted swap partition (with hindsight, the resizing would probably have been quicker if I did this before restoring my data; even so, it was only a few minutes). I didn’t follow the “Install Clevis” section, as I prefer to enter the decryption key before the machine boots. I used the Hibernate Status Button extension to enable the hibernate menu option on the desktop.

Suspend didn’t work properly. The system would suspend, but not wake up. I eventually solved this via this Stackexchange post. You need to add intel_iommu=off to the GRUB_CMDLINE_LINUX_DEFAULT options in /etc/default/grub and then update-grub.

I also took this opportunity to add DISABLE_OS_PROBER=true to the file, so we don’t have the option to boot to Windows from grub, which causes issues with Bitlocker, as mentioned above.

I also had to force an update of the installed Snaps so that Firefox was at the same version as on the 20.04 system, so that it would load my profile data. Presumably this would’ve happend automatically at some point anyway.

It’s built using django and geodjango, with leaflet for the mapping.

With thanks to this post, I figured out a way to do this.

Essentially you download your books using Chrome on the phone, and then use termux, which give a Linux-like environment on the phone, to run a minimal web-server whose output works with the Kobo’s browser.

termux-setup-storage # Gives access to the Download folder on android 
pkg install python3 # Install Python

ifconfig # Note the ip associated with wlan0
cd ~/storage/downloads # Assuming this is where the book is
python -m http.server 8080

Introduction

This project has been a while in the making, and I’ve finally got round to finishing it off (well as much as anything is finished). A big part of the delay was, somewhat ireonically, not having a power socket near the meters.

I’m not a huge fan of Smart Meters - being able to be cut off remotely doesn’t seem a selling point, but I was curious how much energy I was using. I built a circuit with the Raspberry Pi to keep track of much gas and electricity I was using - this uses the GPIO pins to count the flashes on the electricity meter, using an LDR, and to count the rotations of the gas meter using a Hall probe (it turns out the 0 has a magnet in it, which can be detected as it rotates).

The measurements are much more precise for the electic meter - there are 1000 flashes per kWh, compared to ~9 per kWh for the gas meter.

I was originally simply dumping the timestamps (and time difference between successive ones for each power source) into a CSV file, and looking at this in R.

I wanted to make something where I could see at a glance how much energy I’d been using. I had an old Nook E-Reader, which I planed to use as a display. This blog post was very useful in figuring out how to do this.

tldr; it runs Android 2.x, the apks linked to on that page were sideloaded via adb, so no need to set up a Google Account. The plan was to use Electic Sign to display a web page containing the current power usage.

Introduction

I’m currently running a pilot Shiny web hosting service at the University of Manchester. One issue with the standalone (free) Shiny Server is that it’s not particularly easy to deploy apps to it. Essentially you have to copy them to the server (e.g. using scp), and sort out any library dependency issues. RStudio’s Posit’s commercial offerings, like RStudio Posit Connect make this process much easier, but are more expensive¹.

To fill this gap, I wrote an R Package to simplify the deployment of apps to a standalone, open source Shiny Server: Shinysender

install.packages("devtools")  # If you don't already have devtools installed
devtools::install_github("UoMResearchIT/r-shinysender")

Sys.setenv(SHINYSENDER_SERVER="myshinyserver.mawds.co.uk")  
Sys.setenv(SHINYSENDER_USER="david")