I'd love to see something unusual but fun like Sharp SM83 (Game Boy CPU core) π none of that modern RISC-V rubbish...
...and before anybody asks, yes I've put an SM83 inside an FPGA-based SoC design as a control core
Looks like the CDN cache hadn't picked up the update yet so it didn't actually appear...I've purged the cache manually so the feed should now be visible on the site π
I didn't have it but now added an RSS feed. Please let me know if there are any problems with it! I don't write often but when I write, I try to write about meaningful stuff, preferably info that is not available elsewhere. I don't claim to be a good writer but there won't ever be ads or AI slop
New blog post: In case you've ever wondered why Game Boy Advance boards have a 10 nF capacitor on the d-pad up input, but not on the other directions (left, right, down):
gekkio.fi/blog/2026/ga...
> Military: Radar and Sonar
> High-Speed Data Acquisition and Generation
> Video Broadcasting: IP-Based Multi-Format Transcoder
Clearly a special chip for high performance applications!
...SN74LVC1G125 Single Bus Buffer Gate, literally one of the cheapest 1-bit digital buffers from TI
Claude CLI autosuggestion suggests the text "how does To provide five frequently modified core application logic filenames, I would need: work?"
Sometimes you just have one of those days when you ponder the biggest questions in life and want to ask
"how does To provide five frequently modified core application logic filenames, I would need: work?"
On the way to Oxidize conference in Berlin. If you want to chat about Rust, Game Boys, or any other interesting topic, look for a bald (and bold) Finnish guy π
Related: when some operation flashes a spurious error state before loading the real content. How about a loading indicator or even a blank page instead?
For example, my accounting software flashes an alarming AAAGH TRANSACTION CANCELED page for <1s before loading the page for approving/canceling.
One of the most annoying UX behaviours is showing an "empty" result when there's a technical error, and not indicating the error at all.
Spotify search? "lol no results found for X"
Mobile app of my *bank*? "lol you have no accounts"
Accounting software? "lol you have no invoices/receipts/anything"
Isn't it also counting Result values instead of actual lines, i.e. it counts and hides errors instead of propagating them? π€
Considering this, I'd perceive "A gentle nudge toward correctness" as an insult π
Perfect for code reviews: "It doesn't work but there's a gentle nudge towards correctness"
As a user I don't care which programming language something is written in, but experience tells me there's a certain advantage when one vendor is evaluating different tools than just "programming harder and more carefully" and redoing the same mistakes
Maybe I'm weird but the fact that LibrePCB is strategically moving towards Rust is quickly making it more attractive to me. I've been a happy KiCad user for almost 10 years but I've seen my share of crashes, bugs and various "C++ problems".
My opinion about USB-C was forever changed when I recently watched in horror as somebody shoved a USB-C cable into a USB-A port, shorting the pins with the USB-C plug shell.
The idea of having a computer with no USB-A ports and *only* USB-C doesn't seem so silly anymore... π
It's the Apple IIc (not this unit specifically...this photo is from homecomputermuseum.nl). Mine still works but it's in storage at the moment.
It's a cool device and very portable! I remember loading it on a small sled during some winter as a kid, and bringing it with me when visiting a friend.
Age yourself with the first computer you used.
Chisel tips like that are pretty good, but I personally prefer knife tips π Takes a bit of practice to use them well... But in the end it's a personal preference, and the most important thing is to choose a tip that has sufficient thermal mass for larger solder joints (= no tiny "needle" tips)
It's also worth noting that the cartridge uses an MBC2 memory controller and no SRAM chips. It seems very power efficient and keeps the save in a small internal 512x4 bit SRAM. Many games use larger SRAM chips (e.g. 8192x8) that require more current to retain their state, relative to memory size
I desoldered the battery, connected it to a Siglent SDL1020X programmable DC load, and used its battery measurement functionality in constant resistance mode (CR).
Not the perfect setup since the CR mode caps at 10kΞ©, which makes the measured capacity a bit lower than it should be (vs a 30kΞ© load)
Finished testing this *35-year old* CR1616 battery on this Game Boy cartridge.
CR1616s typically have a 60 mAh nominal capacity (with a 30kΞ© load). In my test I got 31 mAh out of this battery with a 10kΞ© load until 2.0V cutoff so the battery had *at least 50%* capacity left after 35 years. Wow!
Thanks TI website for this lovely validation error π
I guess what they really mean is "domain name", not URL (or URL without protocol? π€), but in any case "Please enter the URL in English" is one of the worst validation error messages I've ever seen
Based on a quick look, typst_render seems to return a tiny-skia pixel buffer (RGBA), so you can do extra processing and freely choose the final format π π
It's good for my use case as well, since I need to send the image via USB to a Samsung SPF-87H photo frame, which is a bit weird and wants JPEGs
Ooh, I've got a slightly similar use case and it never occurred to me that I could use Typst π€―
Right now I generate SVG using some pretty dirty text templating, and then render it with resvg. It works but layout with raw SVG is a PITA...
Are you invoking typst-cli or using the lib crate directly?
Screenshot of Github Home page with multiple feed entries. The top entry says "3 weeks ago" and the bottom entry says "2 weeks ago"
Screenshot of Github Home page with two feed entries, scrolled down until the More button appears. The upper entry says "20 hours" and the lower entry says "21 minutes ago"
Mwahaha, somebody at Github pushed to production and messed up the home page a bit π my feed is now sorted oldest to newest for some reason
I did some PCB design today and was affected by *two* part manufacturer website outages on the same day when I needed to find info about some of their parts. These companies are not even some random small ones...
Samtec gave me error 500, and WΓΌrth vomited some aspx errors
Why is it always the big corporations with all the money in the world that have the worst websites that are down exactly when you need them?
WΓΌrth and Samtec, I'm disappointed! π¬
