Reorganizing development work stints in order to make the best use of the available AI credits before they top up the following month.
The trend that no one saw coming.
Reorganizing development work stints in order to make the best use of the available AI credits before they top up the following month.
The trend that no one saw coming.
I cannot find the words to express how much I dislike the file permissions mechanism in Windows.
β
Blue Danube
β
Radetzky March
There we go then. Wishing you all a very happy 2026!
So apparently, if you are a huge software vendor and charge big money for support and consulting, you can sell rubbish software and ignore standards and specs, and then expect that your customers will force their OTHER vendors to adapt to your garbage because that will be much cheaper for them.
Automate repetitive UI tasks.
The Gemini 2.5 Computer Use Model lets you build agents to handle forms, reports, and more. The model is trained with safety features and is now in public preview.
Try it β goo.gle/Computer-Use-Model
OK, this is my first one. Let's build something! #raspberrypi
In the middle, there is an AI-generated picture of a duck. Below it, there is a human-illustrated Duck, looking at the other duck in shock, and saying "Wtf is this?"
AI-generated content is flooding the internet, and we're entering a new era of information overload. Watch our latest video to find out how AI slop affects the internet and why kurzgesagt videos will always remain human-made: https://kgs.link/AISlop
Explanation on F#'s system for applying units of measure to numerals.
I wish Java had this.
π From Beans to Boot, Aspects to AI by Rod Johnson / Juergen Hoeller / @starbuxman.joshlong.com @ Spring I/O 2025
βΆοΈ Video: youtu.be/i_fiHmg1qM0 #springio25
I mean... OK. I can do that.
But also "uv python find --show-version".
Anyway. Perhaps I should read the documentation.
I explained that point at the chat, and then came the first thing similar to a "good" answer: use "uv python find", which will output the path of the python runtime, then pipe it to "xargs -L 1 -I {} {} -V" to execute it and extract the version ("-V").
3. So I asked Gemini (2.5 Flash), and this one told me that "uv python" without any arguments would give me the path of the python executable being used in the virtualenv.
Which was incorrect again, as this only outputs the list of arguments that can be passed to it.
So I explained that 0.8.17 was the latest version, released 5 days ago, and its response was that in such a case I should just use "python --version" without uv (OK, that's true, but I wanted a "uv" option). And again that "uv python list" would indicate the current one in the list (nope, nope).
I told it that "uv python --version" did not exist, and that I was using uv 0.8.17β¦ and its response clarified that, in fact, the "uv python" subcommand had appeared in uv version 0.9.0, so I needed to upgrade π
2. So I asked Claude (Sonnet 4), and again it told me that I should use "uv python --version" (nope) and also that "uv python list" would give me a list of the installed python versions, indicating the currently active one (also nope on this last point).
Interestingly, this explanation included a literal quote from the release notes explaining the reboot, and a link to such release notes where apparently I could verify that point myself.
None of that was real. Version 0.8.17 was released five days ago.
I reiterated that I used version 0.8.17, which is newer than 0.1.0 and, as a response, I obtained a lengthy explanation about how, contrary to what it seemed, 0.1.0 was in fact a newer version than 0.8.17 because the uv project had been rebooted after 0.8.0, back to 0.1.0 π€¨
1. Github Copilot (GPT-4.1) told me it was "uv python --version".
I told it that a "--version" option didn't exist for "uv python", and that I was using uv version 0.8.17. Its response was that this "--version" option appeared in 0.1.0, so I needed to upgrade to that version π€·
I've just asked 3 AIs about the way to obtain, using the "uv" command, which is the version of the python runtime that is being used for the active virtual environment.
I know I can use "python --version", but I was curious to see whether I could do everything with "uv".
SQL was dead.
Then Hadoop got the hype, and we put a SQL layer on top to improve it. Then noSQL got the hype, and we put a SQL layer on top to improve it. Then the Data Lake got the hype, and we put a SQL layer on top to improve it.
I can't wait to see what SQL looks like on top of AI agents.
#java JEP 401 on value classes is submitted : openjdk.org/jeps/401
I keep moving my old "small" OSS projects to a new namespace, and so "op4j" (a fluent API for Java similar to today's streams, that I developed back in 2010) now has a new home. Though in this case I have not modernized the code because the project was deprecated by JDK8.
github.com/arxila/op4j
This touches on two critical points most people miss:
First that most OSS, even popular projects, get no relevant external contributions ever.
Second that maintainers get NOTHING out of their effort besides a fancy line in their CVs that most companies ignore.
OSS is 95% fueled by mere dev pride.
And another one of my "old" libraries gets refreshed and moved to a new namespace: JavaRuntype.
JavaRuntype is a runtime representation of Java's type system.
github.com/arxila/javar...
I've given my old "JavaTuples" library a new life, reimplemented as Java (JDK17) records. And I've also moved it to my new Arxila OSS project organization.
Solo, Pair, Trio, Quartet⦠JavaTuples 2.0 is out!
github.com/arxila/javat...
The AI revolution is slowly turning for me into wasting hours angrily debating with an LLM about class and variable names.
Even the smallest piece of software you want to try is now distributed as a 12-container docker-compose deployment or kubernetes cluster. Everything is so slow.
It's incredible how good we are at making our new shiny supercomputers turn uncomfortable and cumbersome at every step of the tech cycle.
Oooh...what are these shiny new lock icons π€
That's right... Spring Modulith support landed in @intellijidea.com 2025.2.
You don't have to run the test to know whether you are violating any module boundaries. IntelliJ IDEA tells you as you code π₯
Spring Boot 4.0.0-M1 is available now:
spring.io/blog/2025/07...
Hello summer.
#sanxoΓ‘n