Notes for June 26-July 2

This was an eventful week. I am not terribly excited about the next few ones, so I’m going to try to keep this short.

Quite a warm day.

• Personal inbox zero, plus a few other chores.
• Experimented with vim-ai, a very straightforward vim plugin that, after a few tweaks, works quite well inside a-Shell on the iPad.

Fired up the AC in the office, which made it a lot more bearable, and spent a while poking at Proxmox to prep upgrading to 8.0. Ended up doing security upgrades to all of my machines.

With temperatures stabilizing, ran a few errands, including yearly routine medicals that afforded me plenty of time to read up on current LLM tooling and the hype therein.

Since we had a family trip planned, I decided to provision a prepaid eSIM for my iPad mini, which turned into a Zelda-grade side quest.

Out of sympathy for my former colleagues who used to work there, I used the iOS Settings app to try to provision a Truphone data plan, but it failed to activate the eSIM – the dialog just went away and nothing happened. For an hour.

This led me down an inscrutable maze of un-supportive support. Calling phone support tells you to visit the website, the website is useless (with or without a login), and my e-mail plea for help to their “award-winning support staff” also pointed me to the website and was instantly closed. Well, inside of four hours, anyway.

Also, there are apparently three ways to log in to “a Truphone account”, and none of them seem to have a direct relationship to the iPad eSIM provisioning process. I had to reset my password using the iPhone app (which doesn’t work on the iPad), and the only way I could get the eSIM to activate was to pick “Other…” as the carrier, which led the iPad to realize it had a pending eSIM activation and complete the process.

A little birdie tells me this is because the iPad eSIM provisioning flow is a bit of a hack on the Apple side, but I cannot confirm this. What I do know is that the dismal Truphone support experience I effectively never had is inexcusable, so next time I’m going to check if Vodafone has their act together (last I checked they still insisted in printed QR codes, which feels barbaric).

Last day of the fiscal year at work, so I tidied up and popped off to Porto for a couple of days.

Walked 12Km around Porto, which was an utterly unintentional feat given my current state of fitness.

While trying to post more photos, I realized that I was not handling asyncio.TimeoutError in a very specific place in my static site generator.

This isn’t the kind of thing you want to debug 300Km from home, and took me a while to figure out on my phone while traipsing round the city (but yeah, logs worked). Eventually I sat down with my iPad mini, fired up a Remote Desktop session to my GNOME box, and deployed a one-line fix in a few minutes (backoff is great, but you need to anticipate what exceptions to handle).

In the process, I also realized Jump Desktop is now at least an order of magnitude less efficient than the official RD Client, and confirmed that Fedora 37’s xorgxrdp uses less bandwidth to stream a GNOME desktop than my built-from-source Ubuntu/XFCE desktop (iptraf-ng is your friend here).

Back in Lisbon. Spent the afternoon nursing my feet and sunburns as well as doing some homebridge maintenance (I am still using some MQTT-based virtual devices to make up for the fact that homebridge-z2m cannot convert contact sensors to motion ones).