▲ 148 points
back
138 comments
'PowerPC DRC' appears to be the code in mame that translates PowerPC machine code into native host machine code (Dynamic Recompilation Core).
https://github.com/mamedev/mame/blob/master/src/devices/cpu/...
Rapid development of a cycle accurate Intellivision emulator, much of it unsupervised. (four part series)
https://keanw.com/2026/03/a-diary-of-an-agentic-retro-gamer-...
https://keanw.com/2026/04/a-diary-of-an-agentic-retro-gamer-...
https://keanw.com/2026/04/a-diary-of-an-agentic-retro-gamer-...
https://keanw.com/2026/05/a-diary-of-an-agentic-retro-gamer-...
some vague hint about what the article is about before clicking on it would be.. nice.
For those unaware, this line is from the movie Terminator 2:
I clicked on the post expecting it to be something from T2 and wondered why I was reading something about emulation.
I have strong affinity for having references popping into my mind when doing something not obviously related. Sometimes the connection is obscure, sometimes the source of the reference is obscure. I highly doubt people would get many of them.
One of my favorite ones is "Third turtle lies.". When people wonder how something reported could have been possible.
When I saw the headline I immediately knew the reference. Then I skimmed that article and saw no connection to T2?
That movie has aged incredibly well!
As a story, yes. But Terminator failed on a basic premise: Skynet becoming self-aware.
The future seems more like Blindsight [1]: hyper-intelligent, completely unconscious systems outperform, out-manipulate, and out-compete human beings purely through automated efficiency.
> That movie has aged incredibly well!
Except for the titular event not happening!
Resolution-wise it hasn't due to the extensive use of early CGI.
I thought it was about Renaud' song _Laisse béton_
My brain immediately switched to 'you could be mine' by guns n roses
I am doing a very similar thing right now, which is adding a JIT to both Basilisk and SheepShaver: https://github.com/rcarmo/macemu - on the PPC side, we're still poking at some hardware registers that the Quadra ROMs like to noodle in.
The blog mentions a Graphing Calculator. Not sure if it shares code, but macOS still ships with an app to draw graphs, Grapher.app
The app on your Mac today isn't a rewrite of the legendary Mac OS 7.2.1 Graphing Calculator, but an acquired app based on Curvus Pro introduced in OSX Tiger.
The first one has a legendary backstory. 2 devs snuck into Apple after their project was canceled: https://www.pacifict.com/Story/
Curvus Pro: https://macintoshgarden.org/apps/curvus-pro
Thanks for sharing the links. Love this part!
“We looked at each other, took a deep breath, and launched the application. The monitor burst into flames. We calmly carried it outside to avoid setting off smoke detectors, plugged in another monitor, and tried again.”
This highlights what I think is missing within a lot of tech companies today. I don’t know too many people who are so passionate about what they’re doing that they would sneak into their former employer’s office buildings and then find employees willing to work together with them to finish a project that simply had to get done. That alone makes me a bit sad.
Just wow! I’m speechless! Can’t think of any other word than “legendary”.
Grapher.app is different from Graphing Calculator. It came via an acquisition. All the details are here if you want to read the backstory (assuming the info is correct: https://en.wikipedia.org/wiki/Grapher)
The current Mac version of what started as the PowerPC Graphing Calculator app can be downloaded from the Mac App Store: https://apps.apple.com/us/app/graphing-calculator-4/id522175...
The iOS version is at: https://itunes.apple.com/us/app/pacific-tech-graphing-calcul...
"I does not boot and it makes me sad"
I actually write prompts like that when I'm not under pressure. Claude will sometimes completely ignore your feelings, and sometimes give a little comment, which I just find refreshing in the middle of otherwise often boring sessions. And it does not have an effect on the actual result.
Codex overuses the word "quickly". I'm tempted to check what happens if I tell him to do it slowly.
Crazy theory: Maybe Codex watched too many YouTube videos. I have noticed on YouTube that many younger narrators say, "I quickly did X," when "I did X" would have been more just as well and usually more truthful. Why did you have to _quickly_ cut that piece of wood when the video clearly shows cutting it at a perfectly normal speed?
Also, please try not to anthropomorphize LLMs, they hate it when you do that.
403 forbidden
Feels like it might be pertinent to share more details than simply the error code. What country? How are you connecting? Anything out of the ordinary with your setup that might be the cause?
Singapore - 223.119.20.232
I also get 403 Forbidden, from EU, nothing special about my setup
I also see that error code, from Europe.
Perhaps it blocks any non-USA connection.
Works for me fine
"You forgot to say please"
A perfect example of a coding agent guided by a human with domain knowledge.
Intuition + obsessive detail-oriented execution = results
you forgot to say 'please'
It's bittersweet, isn't it. Software is solved, but at a terrible cost.
How is it solved? LLMs cannot think new things, they can only cobble something together if it's in their training set.
New things are made by cobbling together existing things.
They can think and reason better than most humans. Most problems they're pointed at are not in their training set, but in certain ways they resemble things that are—maybe there are a few different resemblances to different problems in its training set—so it's able to pull these disparate similarities together and apply the patterns it finds to come up with a solution. Much like human brains do.
What? This is a massive misunderstanding. It’s easy to get truly novel ideas from LLMs, unless your definition of “new things” is so strict that no human can do so either.
The training set is about patterns, not facts or specific configurations. Yes, it’s possible to extract (some) of the training set verbatim, but that doesn’t mean it’s all you can do.
I am not sure if claude had powerpc scripts in its training.
This reminds me of the Go champion who announced he was giving up the game after a computer beat him.
It’s as if a runner were to give up running when beaten by a horse or a car. It suggests they may have had unexamined and perhaps somewhat strange reasons for doing the activity in the first place.
People have difficulty accepting just how significant their limitations actually are. We design our world to hide those limitations. As an example, it would be easy to make computer games that are unwinnable by humans because of our slow reaction times, low speed in general, and our cognitive limitations. But no-one makes such games, because few people would want to play them for very long.
The “terrible cost” in this specific case seems to be related to discovering that we were fooling ourselves about how good we were at software development.
Why is it bittersweet? Carpenters probably didn't cry when their tools improved.
It will be bittersweet when there's no human needed at the wheel but IMHO we are far, far from that. These models/agents are just mimicking human text and need guidance because they often get lost or stuck.
Software isn't solved. 'Coding' is, according to the people of Claude.
Coding (programming) is a tedious and expensive part of software engineering. There's other parts AI isn't doing, such as understanding and refining requirements, and delivery + accountability.
The cost is not terrible, calm down.
Asta la vista baby
Hasta* (it’s spanish)