Activity Log
hackers.town: 2024-12-05 Thu 00:26
Adding white lines to go with the black lines in the bedtime #drawing
hackers.town: 2024-12-05 Thu 10:41
@jalefkowit well hey at least it’s a former CEO.
hackers.town: 2024-12-05 Thu 11:08
@IamMrsGeek it smelled nice, that’s for sure. Just neurospicy brain going “it’s breakfast and today’s breakfast food is banana wrap.”
hackers.town: 2024-12-05 Thu 11:16
Looking at a microbenchmark collection – those are always suspect, but still fun to peruse– and even before I get to the actual code I have thoughts..
The run.sh launcher script has a lot of what I can only describe as “run this code in that language’s debug mode,” which at best is not a useful bookmark.
That is such low-hanging fruit that I’m tempted to fork and adjust.
hackers.town: 2024-12-05 Thu 11:44
Though really, this issue puts the benchmark’s problems most succinctly:
“yet another microbenchmark suite that doesn’t measure anything useful and does it in the most inaccurate way”
Replace entire suite with one that has already been golfed to death · Issue 157 · bddickenlanguages
hackers.town: 2024-12-05 Thu 11:56
https://hackers.town/users/earthshine/statuses/113601733972245586
hackers.town: 2024-12-05 Thu 12:19
I’m not hugely fond of the Bay Area on the whole but I’d strongly prefer everybody in it stay safe and dry.
hackers.town: 2024-12-05 Thu 13:24
So. I went ahead and tried the flawed benchmark suite just for curiosity. The author shared results generated on their M3 MBP w/16GB RAM. My M1 MBP w/32GB was mostly ~25% slower (which makes sense to me, more RAM offsetting slower processor).
Except Python. Author got 75s, I got 46s. WTF? Oh they used 3.9 (which was demoted to security fixes in 2022. I used 3.13.
Anyways, no I’m not going to fix the code or invocations. Off to look at some real benchmarks.
The misleading visualizations were neat though.
hackers.town: 2024-12-05 Thu 13:40
@cinebox bad benchmarking. Goes off system clock, so any background processes mess with results.
Also, I’ve got a significantly slower processor but double the RAM, so that skews things.
Finally, the benchmarked code is itself pretty naive. Wouldn’t be surprised if it works against Python version optimizations.
hackers.town: 2024-12-05 Thu 14:33
Off to get ADHD meds with the GoodRx coupon thing instead of insurance. Let’s hope this is minimally problematic.
hackers.town: 2024-12-05 Thu 15:27
All resolved, no drama, picking up actual refills tomorrow.
hackers.town: 2024-12-05 Thu 22:10
@annaleen @evan Unamerican!
The site, that is. He (or somebody else associated with the site? quarter-century old memory is fuzzy) came into the gas station I was working at immediately before officially in tech.
Recognized his shirt from unamerican.com, blurted “Unamerican!” had a very quick chat and got gifted some “fuck work” stickers to place strategically throughout the gas station.
Glad he’s doing okay. He (or his employee?) seemed nice.
hackers.town: 2024-12-05 Thu 22:14
@diffractie the feeling in my brain reading that phrase is like how my fillings feel driving on Arizona highways.
(replace “Arizona” with whatever place you’ve been at with terribly maintained roads)