Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
breck
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
51 ms
·
751.
▲
by
breck
2y ago
https://web.archive.org/web/20240512231249/http://cpudb.stan...
752.
▲
by
breck
2y ago
We classify JSON as a "dataNotation" (line 4 on https://github.com/breck7/pldb/blob/main/concepts/json.scrol... ) We classify something like C as a "pl" (programming language) (li
753.
▲
by
breck
2y ago
Yes. A small (but extremely popular and useful) domain specific language for describing arrays and objects and scalars.
754.
▲
by
breck
2y ago
Having worked on languages and compilers for many years, I thoroughly enjoyed this post. Another way to phrase the same idea: a language is _a lot more_ than just compile speed. Compile speed is very important, in the top 10 dimensions, f
755.
▲
A Language Without Comments
(pldb.io)
3 points
by
breck
2y ago
|
5 comments
756.
▲
by
breck
2y ago
Hadn't heard of this ( https://kdenlive.org/en/ ). Thank you!
757.
▲
Ask HN: Is there a single website with power law impact on trained LLM?
2 points
by
breck
2y ago
|
0 comments
758.
▲
by
breck
2y ago
I struggled to use it but it seems like it has potential and I starred it. I love how you land us right into the editing experience. Maybe there's a way to add some kind of short fully featured demo video when a fresh visitor lands? I
759.
▲
The Office: Words Spoken by Episode
(twitter.com)
1 points
by
breck
2y ago
|
0 comments
760.
▲
Marks
(breckyunits.com)
2 points
by
breck
2y ago
|
0 comments
761.
▲
by
breck
2y ago
Nice read. (Any chance you could add an RSS feed or similar?)
762.
▲
by
breck
2y ago
Well put. Also, ever read your old journals? You are training on generated data.
763.
▲
by
breck
2y ago
I tried to fetch his RSS too! :) Turns out, there's only 1 post so far on his blog. Hoping for more! This one is great.
764.
▲
The Box Method of Science
(breckyunits.com)
1 points
by
breck
2y ago
|
0 comments
765.
▲
Apple Is Developing AI Chips for Data Centers, Seeking Edge in Arms Race
(wsj.com)
44 points
by
breck
2y ago
|
53 comments
766.
▲
QUAMELEON: QUAntum Materials Engineering using eLEctrOmagNetic fields
(darpa.mil)
14 points
by
breck
2y ago
|
2 comments
767.
▲
30k Hours (2019)
(breckyunits.com)
2 points
by
breck
2y ago
|
0 comments
768.
▲
by
breck
2y ago
Update: I've reverted back to Google for now. Kagi was neat, but requiring login for search was very bad ux(for example, I use multiple browsers, incognito mode, etc). I like the Google product, it's the search index that's f
769.
▲
Three Scientists Go to Heaven
(breckyunits.com)
7 points
by
breck
2y ago
|
3 comments
770.
▲
by
breck
2y ago
Ah, hadn't tried that yet. Pretty neat. Will go in my toolbox, thanks. Still a little too commercial to be my primary, but thank you!
771.
▲
Ask HN: A search engine that excludes all sites with ads and/or paywalls?
4 points
by
breck
2y ago
|
8 comments
772.
▲
by
breck
2y ago
> The third step is to figure out what went wrong and what's needed to fix it. Very much agree. Lately I've been wondering if depression is nature's way of forcing us to leave behind our current ways of thinking and grow a
773.
▲
by
breck
2y ago
This is one of my favorite SW essays. A tour de force in terms of using CA to show evolution, mutation, and natural selection. Don't miss the "Personal Notes" section at the bottom. Neat history and insight into the process a
774.
▲
by
breck
2y ago
Perhaps the root explanation of why he writes about himself so much is that he is a passionate observationist/scientist. He observes a lot of the things around him, measures these things, things about their underlying models, and write
775.
▲
Wearable Pass Through Screen (1962)
(twitter.com)
2 points
by
breck
2y ago
|
0 comments
776.
▲
by
breck
2y ago
Paper: https://arxiv.org/pdf/2404.05699
777.
▲
Image shows atoms transforming into quantum waves
(livescience.com)
4 points
by
breck
2y ago
|
1 comments
778.
▲
Bloggers Turning Pro (2008)
(breckyunits.com)
1 points
by
breck
3y ago
|
0 comments
779.
▲
The Three Byte Fix
(breckyunits.com)
59 points
by
breck
3y ago
|
5 comments
780.
▲
by
breck
3y ago
> only believed to be faster than classical computers for a handful of very specific problems But aren't these problems all in the factorial and exponential classes?
More ›