Kinrany

5 ms·

61.
by Kinrany 2mo ago
> No taxpayers should have ever been allowed to use it to file a tax return. I see why blowing too much money on the study is a problem. Though this needs to be compared against the norms around this kind of thing. If we wanted these lim
62.
by Kinrany 2mo ago
You can let it help with syntax, but you can't rely on it to understand precise semantics. SQL is the worst offender here: LLMs are very good at remembering its awful syntax, but very bad at writing nontrivial queries.
63.
by Kinrany 2mo ago
Tests not being core seems suspect
64.
by Kinrany 2mo ago
Like tests, it should only be used as a safety net and a signal that shows whether the way the program is written is prone to errors
65.
by Kinrany 2mo ago
As far as I understand flight numbers are reused regardless, they're more like bus lines
66.
by Kinrany 2mo ago
This seems backwards. We've known for a long time that successful execution is not enough for understanding.
67.
by Kinrany 2mo ago
You need to actually remove the bottleneck, after _that_ there'll be a new one
68.
by Kinrany 2mo ago
Funny. I love flying partly for the very reason that airports need to be idiot-proof.
69.
by Kinrany 2mo ago
They might still be available for killing someone out of gratitude
70.
GitHub Stacks in Jujutsu(alan.norbauer.com)
2 points by Kinrany 2mo ago|0 comments
71.
by Kinrany 2mo ago
I wonder why operating systems aren't integrating with browsers more.
72.
by Kinrany 2mo ago
As a fellow senior professional I too keep vacuum tubes in my toolbox.
73.
by Kinrany 2mo ago
"Creating value", "making money", etc. are meaningless
74.
by Kinrany 2mo ago
That's not an interesting question, it'll be the same path as before but somewhat longer.
75.
by Kinrany 2mo ago
If you're writing CRUD, there's no excuse not to write it competently since you're not solving a new problem
76.
by Kinrany 2mo ago
For example, doing a marketing push assuming that the app works, and then having it fail in trivial ways at release.
77.
by Kinrany 2mo ago
I believe compressed random noise cannot be shorter in expectation than the original
78.
by Kinrany 2mo ago
Is it a goal to actually run self-hosted runners against preloop?
79.
by Kinrany 2mo ago
Every single thing people care about is about people when discussed by people
80.
by Kinrany 2mo ago
You mean the thing that was wrong with DOGE was execution, not the idea itself
81.
by Kinrany 2mo ago
I don't follow, how is that better? It's down more often AND you have to do more work.
82.
by Kinrany 2mo ago
Wait, were Webhooks also down, not notifying about commit pushes?
83.
by Kinrany 2mo ago
Yeah these results are useless when `npm run setup` is "dangerous" but then `npm run lint` somehow isn't. Not only do these tests lack context about the execution environment, they're straight up not even self-consistent
84.
by Kinrany 2mo ago
It doesn't matter if the results are bad even when the devs know that it's a test.
85.
by Kinrany 2mo ago
~scarcity~ prices //_-
86.
by Kinrany 2mo ago
They are already trained on generated data I believe
87.
by Kinrany 2mo ago
We have a perfectly good mechanism for dealing with scarcity, it's called scarcity. It's more plausible that the issue is this being a lemon market
88.
by Kinrany 2mo ago
The quest for Torment Nexus continues
89.
by Kinrany 3mo ago
Yeah: they copied what people were building on top of Github as a workaround to not being able to change Github
90.
by Kinrany 3mo ago
Jujutsu makes all this easy

More ›