Is it still the norm to go to the dev’s office, yank their power cord and when they ask what we’re doing, tell them we’re shipping their machine to the client because it’s the only one that the code runs on?
And can we do that with whatever server ChatGPT-4o is running on?
I’m assuming that this response from 4o isn’t real and was invented for the laugh, but it would be tempting to throw this scenario at it if it decided to give this response.
The real joke was the pain that every developer feels when the end user gives such useful and actionable feed back as “It broke. Fix. Unga bunga.”
“It works on my machine” is trying to be polite when, after hours and hours of trying to teach a person how to report a bug with necessary information, all they ever get is “It broke. Fix. Unga bunga.”
You can’t deny that it correctly predicted the most likely token in this case.
That’s why
dockerwas created.Imma let you finish but Nix had the best repeatable, declarative dependency management of all times…of all times.
Is docker even declarative?
Also you can build docker images from nix derivations
Is docker even declarative?
Yes (though not as much as Nix).
Also you can build docker images from nix derivations
Yes. I know.
1950s
A: The transistor I made using your blueprint doesn’t switch properly at 12V.
Maker of Blueprint: The one I made, works at 12V.why?
Blueprint was made by a person in the tropics.
A was in Europe
You’re probably using the wrong compiler flags, did you remember
-Wall -Oz -nostdlib?-Oz
Optimize aggressively for size rather than speed.
TIL
It is, honestly, the dumbest of the
-Oflag option, which is why I picked it. I’m sure there are times when it’s useful, but it’s nearly never the right choice.Wasm comes to mind, execution time in the browser will probably be ok, but size is a big deal
Software that runs on embedded systems usually benefits from being small, too.
As someone who has worked on embedded systems for the past 30 years: It used to be a real big deal, but for the past 10-15 years it hasn’t. We now have fully fledged multi core systems running everything. Even small embedded sensors or actuation controllers are 100+ MHz microcontrollers with oodles of flash and ram.
Now there has been an interesting turnaround with the whole chip shortage for the past years. All the young folk are at a loss, being used to just putting powerful chips all around willy-nilly. So they turn to the old folk like me to figure out designs with less chips, running busses all over and connecting dumb sensors/actuators to a central processing unit.
You can’t deny that it correctly predicted the most likely token in this case.
“an error” okaay
Damn, maybe we can be replaced…
These are exactly the people it will replace.
The question is, which one will write shittier code that the rest of us need to clean up.
Learned from the best
