blueybingo

5 ms·

1.
by blueybingo 5mo ago
wasn't this basically the consensus among numerical analysts like 20 years ago? i remeber reading similar arguments in goldberg's paper and various game dev forums circa 2005, so genuinely curious what keeps making this idea feel
2.
by blueybingo 5mo ago
damn brain is becoming bot mush
3.
by blueybingo 5mo ago
but why does "the industry ignored it" hold as the central framing when the actual story seems to be "the DoD mandated it, contractors used it, and it worked fine for exactly what it was built for"? the implicit assumpti
4.
by blueybingo 5mo ago
haha no im not a bot, but starting to realise i sound like one. need to be less cynical.
5.
by blueybingo 5mo ago
the article glosses over something worth pausing on: the `getattr` trick for dispatching instructions (replacing the big if-elif chain) is actaully a really elegant pattern that shows up in a lot of real interpreters and command dispatchers
6.
by blueybingo 5mo ago
worth noting that the google stat measures ipv6 availability among users who access google, not general internet traffic -- so it's a bit of a self-selecting sample skewed toward consumer isps that have deployed ipv6, which probaly o
7.
by blueybingo 5mo ago
the zero injection fix for sparse counters is the most underrated part of this writeup -- injecting a synthetic zero on first flush to anchor the cumulative baseline is actaully a pretty elegant solution to a problem that bites almost every
8.
by blueybingo 5mo ago
the article buries what's actaully the most practical gotcha: ollama's hashed blob storage means if you've been pulling models for months, switching tools requires re-downloading everything because you can't just point a
9.
by blueybingo 5mo ago
the article's framing around nanopass is undersold: the real insight isn't the number of passes but that each pass has an explicit input and output language, which forces you to think about what invariants hold at each stage. that
10.
by blueybingo 5mo ago
the missing piece in most normalization discussions is the OLAP vs OLTP split. in analytical dbs denormalization isnt a mistake its a deliberate tradeoff for scan performance. teaching normal forms without that context sets people up to mak
11.
The three villains to agentic observability: retention, sampling and rollups(clickhouse.com)
3 points by blueybingo 5mo ago|0 comments
12.
by blueybingo 10mo ago
Also mit and apache if you’re not fond of agpl
13.
by blueybingo 10mo ago
Clickhouse, with their clickstack stack, for logging and traces - it’s not there for metrics yet but great for the other two.