Our main solar install has now been active for three years.
Although the beginning of this year under performed compared to the same time in 2025, we're currently in the middle of a heatwave and the pounding sun has driven yields up at the last minute.
This post is a quick look at our solar performance over the last 12 months.
My Saab recently started to throw a warning when first starting the car:
To clear the warning, I had to turn the ignition off and on again - the warning didn't fire the second time.
The warning itself is triggered by the battery's voltage dipping too low under the demand of the starter.
Just occasionally this led to a failed start and I'd need to jump the car to be able to get it started.
Normally, this would point towards a knackered battery or the alternator failing to charge the battery properly. However, I replaced the alternator relatively recently, and the battery in the car was only about 4 months old (the one the preceded it was knackered though).
This post describes troubleshooting and fixing the issue. It should also apply to the Vauxhall Vectra (and the basic theory applies to most cars).
Over the last few years, I've put quite a bit of work into our solar generation stats, tracking generated electricity value and calculating when the install will break even.
The calculations marry each kWh generated with the Octopus Agile import cost at the time, using an additional calculation to account for the lost export value of anything charged into the battery.
The logic used is quite complex but, in theory, generates an accurate representation of generated value.
However, it is also fundamentally flawed: if we didn't have solar (and, in particular, the battery) we wouldn't be using Octopus Agile.
Prices change every 30 minutes and can be unbelievably cheap (negative even) or preposterously expensive - the whole point in it is load shifting, without a battery peak times would be ruinous.
This means that the numbers don't tell us very much about what we're saving compared to not having the system at all. So, I decided to build a simple report which calculates the value of solar generation based on the average fixed tariff price.
This post talks about that report and how the results compare to my original calculations.
Recently, I've seen posts by a couple of people about adding human.json to their sites.
At it's core, human.json is a simple attestation that the content they publish is written by a human and not by AI.
Obviously, any AI generated site could also publish a human.json, so the system works as a web of trust - the user trusts Alice, who states that Bob is a human, allowing the user to also trust Bob.
There are a couple of browser addons (for Firefox and for Chrome) which can display an indicator of humanity.
This short post covers creating a script to easily add human.json generation to the build workflow of my sites.
Our solar install is not far off three years old.
We've got two east facing panel arrays providing 3.2kWp of PV capacity, supported by a 6kWh battery.
However, only having an easterly exposure has always been a little bit of a bugbear. Although the panels work throughout the day, they start to wane around lunch time, wasting some of the potential that a sunny afternoon can bring.
The house itself doesn't have a usable roof with a westerly (or southerly) exposure. The garage, however, does.
Although it obviously won't generate as much as the existing install, I decided to order a balcony solar kit for the garage, so that we could mop up some of the afternoon sun.
This post describes the install process along with the initial results (despite the weather being a bit... British).
The length of my posts can vary dramatically - some (like this one) are very short, whilst others are much, much longer.
Although I generally try to indicate how long a post is, once a reader is in, there isn't really a good way to know how much more of the post is left (the days of thick scrollbars are, thankfully, quite some way behind us).
So, I decided to add a small visual indicator - a coloured bar which creeps across the top of the page as you progress.
Most functionality on my site works without javascript so, ideally, I wanted the new indicator to be CSS only.
This post describes how to add a progress bar to a website without using javascript.
A few weeks ago, I migrated our e-book library to Booklore. Unfortunately, it seems that I backed the wrong horse: Booklore has since ceased to exist.
Unfortunately, toxicity isn't uncommon in open source projects, so I'm not going to add to it by talking much about the context behind Booklore being deleted.
However, as it's relevant to a decision that I made in this process, I will say that I was already a little concerned about the ongoing quality of Booklore, so it doesn't surprise me too much to hear that other people were being quite vociferous about the developer's use of AI coding tools (amongst other, IMO bigger, issues).
Whatever the rights or wrongs of it, Booklore's repo no longer exists, having been deleted along with the project's Discord and website.
There is a fork (called Grimmory) that I could have switched to. However, given that I already had some quality concerns with Booklore, I decided that I didn't want to gamble on the new fork's ability to clear through tech debt. I'll probably check back in six months to see how things have progressed though.
Instead, I decided to move to Calibre-Web Automated (CWA). It's a fork of Calibre-web with some additional functionality like Smart shelves and KOReader sync support.
Both Calibre-Web and CWA consume an existing Calibre format library.
Although I still had my original Calibre library, I'd since added a number of new books to Booklore. Rather than risking mistakes by manually identifying and adding them, I decided to migrate the entire library from Booklore into Calibre-Web as if I'd simply had a filesystem directory full of ebooks.
Earlier this week, Cloudflare announced the introduction of their Browser Crawl Endpoint.
This allows Cloudflare users to crawl an entire website by making a single API call to the Browser rendering service.
Although the browser rendering service honours robots.txt they don't define a specific User-Agent that the service will check for, apparently instead expecting website operators to disallow all user agents if they want to keep Cloudflare out.
However, they have also documented that the service includes Cloudflare specific request headers, allowing requests to be blocked by checking for those.
This post details how to achieve that on BunnyCDN, Nginx and Openresty.
This week, I set up a Booklore instance to allow me to wirelessly sync ebooks to my Kobo e-reader.
However, in the (short) time since, I've run into a bit of a limitation: Booklore's Kobo sync doesn't support PDF files (which seems to be a limitation of the way that Kobo fetches files from its store rather than Booklore specific).
Most of the books that I read are in EPUB format, but I've also got various user manuals and technical reference docs in PDF that I'd like to be able to fetch onto the reader, on demand.
Although I could use a USB cable to sync PDFs over, it'd somewhat undermine the point in running Booklore in the first place.
So, I decided to install KOReader onto my Kobo.
KOReader is a document reader designed for e-ink devices and can fetch books (and more importantly, PDFs) from Booklore via OPDS, syncing reading progress back via Booklore's dedicated KOReader api.
This documentation details how to install KOReader onto a Kobo before configuring it to work with Booklore. If you're a Calibre user, KOReader can also talk to Calibre.
Last weekend, a blog post about Booklore crossed my feed and piqued my interest.
I've long used Calibre to manage my e-books and, in June last year, even containerised Calibre-server so that I could read (parts of) my library in a web browser.
As fantastic as Calibre is, its interface is showing its age and I don't love that they've been diverting energy into adding AI features. I'm sure that they're probably useful to someone, but I tend to view reading as an escape rather than an opportunity to further immerse myself in industry hype.
Booklore though, looks shiny and new.
One of its features - Kobo sync support - also caught my eye. This masquerades as the Kobo store, allowing physical e-readers to fetch books remotely.
This post talks about getting Booklore up and running and playing around with it - it details some of the niggles that I ran into as well as some of the tweaks that I've made.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。
