Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
tromp
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
50 ms
·
601.
▲
by
tromp
2y ago
https://archive.is/yeoJI
602.
▲
by
tromp
2y ago
https://archive.is/7LsG3
603.
▲
by
tromp
2y ago
Recalls come in two types, soft recalls that can be fixed by a software update, and hard recalls that cannot. Many people are still not used to regarding the former as recalls, as they associate the word not with the safety issue itself but
604.
▲
by
tromp
2y ago
Counting also starts at 0. The reason many people believe it to start at 1 is due to them calling out counts only post-increment. For instance, if during a car ride you play a game of counting the number of pink cars you see along the way,
605.
▲
by
tromp
2y ago
https://archive.is/jhigW
606.
▲
Donald Trump and Tulsi Gabbard are coming for the spooks
(economist.com)
7 points
by
tromp
2y ago
|
3 comments
607.
▲
by
tromp
2y ago
https://archive.is/nEvr8
608.
▲
by
tromp
2y ago
> This program writes "Hello World" (without quotes) to the output. > This program outputs Hello World. So that alternative doesn't quite work for writing "its own source" without quotes...
609.
▲
by
tromp
2y ago
https://archive.is/yLYGv
610.
▲
by
tromp
2y ago
https://archive.is/tLPXI
611.
▲
by
tromp
2y ago
The ChessPositionRanking project, with help from the Texel chess engine author, tries to prove random positions (that are not obviously illegal) legal by constructing a game ending in the position. If that fails it tries to prove the positi
612.
▲
by
tromp
2y ago
But all such triples are non-primitive. I.e. they are of the form (ka, kb, kc) with k>1. So all Pythagorean triples can be generated as a scaled primitive triple (a,b,c) that is generated by Euclid's formula from m,n.
613.
▲
by
tromp
2y ago
There is one birthday that is far less likely than others, courtesy of leap years. All other are roughly similar, with small seasonal variation.
614.
▲
by
tromp
2y ago
> For one, gpt-3.5-turbo-instruct rarely suggests illegal moves, even in the late game. This requires “understanding” chess. Here's one way to test whether it really understands chess. Make it play the next move in 1000 random legal
615.
▲
by
tromp
2y ago
https://archive.is/WUEv9
616.
▲
Vladimir Putin is in a painful economic bind
(economist.com)
7 points
by
tromp
2y ago
|
3 comments
617.
▲
by
tromp
2y ago
You mean when the guy says "It's intense" ? I was hoping there'd be a 10 second window or so where I can listen to the sound without any distractions.
618.
▲
by
tromp
2y ago
Where is this over half hour long video can I hear the sound?
619.
▲
by
tromp
2y ago
I think fixed supply is overrated: https://john-tromp.medium.com/a-case-for-using-soft-total-su...
620.
▲
by
tromp
2y ago
Funny how correct is not associated with right. The normal association is reflected in he Haskell Either datatype for describing computations that either run into some error (Left error) or run successfully producing a value (Right value).
621.
▲
by
tromp
2y ago
https://archive.is/YHdJm I listened to some of their music but found it rather dull. (see https://www.youtube.com/playlist?list=PLi7orXRWdur4swF3YhpXx... for my taste of music.)
622.
▲
by
tromp
2y ago
> whose hash codes are crypographically (slowly) calculated That's a rather clumsy way of saying that blockchains have Proof of Work. In combination with the Longest Chain Rule (which actually means heaviest rather than longest) th
623.
▲
by
tromp
2y ago
> Another approach has been to design hashes targeting a middle ground; these would have some security requirements but maintain the speed and efficiency of a non-cryptographic hash—for example, SipHash. Sadly this hash function was not
624.
▲
by
tromp
2y ago
There's a somewhat similar Android app at https://play.google.com/store/apps/details?id=com.fortytwo.h...
625.
▲
by
tromp
2y ago
I bet many "elders" (not even retirement age) would line up to get "abused" to the tune of $20M. The outcome was not at all surprising but the payouts were.
626.
▲
by
tromp
2y ago
https://archive.is/Vq4ZO
627.
▲
New Google Gemini model goes straight to the top of the LLM leaderboard
(tomsguide.com)
2 points
by
tromp
2y ago
|
0 comments
628.
▲
by
tromp
2y ago
I would prefer a standard list method/function to test for emptiness, which would be both readable and efficient.
629.
▲
by
tromp
2y ago
> After all, so long as the popular and electoral college votes matched, it doesn’t really matter what system you’re using, and there’s better things to worry about most of the time. It could still matter, since the electoral college wou
630.
▲
by
tromp
2y ago
That's not Haskell AFAIK. Do you mean a >>= b also known as monadic bind of a and b?
More ›