rockwotj

6 ms·

31.
by rockwotj 8mo ago
Technology does not determine the success of a company. I’ve seen amazing tech fail, and things strapped together with ducktape and bubblegum be a wild success.
32.
by rockwotj 8mo ago
https://rockwotj.com/blog/async-wasm-in-seastar/ - been meaning to write about this for a long time. Thanks for the nudge :)
33.
by rockwotj 8mo ago
I have done this integrating wasmtime into a a C++ seastar.io application. I’ll have to write a post about it
34.
by rockwotj 8mo ago
Claude is king for agentic workflows right now because it’s amazing at tool calling and following instructions well (among other things)
35.
by rockwotj 8mo ago
I thought everyone was just using open telemetry traces for this? This is just a classic observability problem that isn’t unique with agents. More important yes, but not unique functionally.
36.
by rockwotj 8mo ago
networking costs are so high in AWS I doubt this makes sense
37.
by rockwotj 9mo ago
It’s so hard to actually benchmark languages because it so much depends on the dataset, I am pretty sure with simdjson and some tricks I could write C++ (or Rust) that could top the leaderboard (see some of the techniques from the billion r
38.
by rockwotj 9mo ago
With the new GC I really love the trend of understanding memory bandwidth is the bottleneck for many things and the combination of locality and SIMD is a big performance unlock. Reminds me of the WAND vs MAXSCORE discussion by turbopuffer:
39.
by rockwotj 9mo ago
you also get a very slimmed down interface that is usually way faster to load. one of the reasons I love HN is that it is super snappy to load and isn’t riddled with dependencies that take forever to load and display. Snappy UIs are always
40.
by rockwotj 10mo ago
Isn’t codex open source and you can just go read what they do? I have read the gemini source and it’s a pretty simple prompt to summarize everything when the context window is full
41.
by rockwotj 10mo ago
Thankfully all these LLM labs are heavily invested in python so this seems like the likely route IMO
42.
by rockwotj 10mo ago
this isn’t the right way to look at it. It’s really server side rendering where the LLM is doing the markup language generation instead of a template. The custom UI is usually higher level. Airbnb has been doing this for years: https:&#x2F
43.
by rockwotj 10mo ago
Redpanda is https://jepsen.io/analyses/redpanda-21.10.1
44.
by rockwotj 11mo ago
Checkout redpanda
45.
by rockwotj 11mo ago
Porting LevelDB[1] to Seastar[2], for internal metadata storage in Redpanda[3]. Before you ask why can’t something off the shelf be used, seastar has unique constraints around its runtime and its memory allocator that means we can’t reuse a
46.
by rockwotj 11mo ago
yes thanks
47.
by rockwotj 11mo ago
> Direct I/O means no more fsync: no more complexity via background flushes and optimal scheduling of syncs. There's no kernel overhead from copying and coalescing. It essentially provides the performance, control, and simplici
48.
by rockwotj 1y ago
I would love to use gdb through an agent instead of directly. I spend so much time looking up commands and I sometimes skip things because I get impatient stepping over the next thing
49.
by rockwotj 1y ago
Also move_pages is about numa nodes not spilling to swap
50.
by rockwotj 1y ago
That seems to be contradictory, as they talk about using it to spill to disk in a simpler way. You can’t swap with mlock?
51.
by rockwotj 1y ago
For bazel: https://github.com/aherrmann/rules_zig Real world projects like ZML uses it: https://github.com/zml/zml
52.
by rockwotj 1y ago
Yeah I mean if you have a latency sensitive workload, you don’t want page faults and swapping to give hidden latency spikes - it kills your P99 latencies
53.
by rockwotj 1y ago
Yeah I built a custom sysroot for Redpanda (Bazel/C++/Distributed Kafka) using a really simple docker image: https://github.com/redpanda-data/redpanda/blob/dev/bazel/too... We also have an
54.
by rockwotj 1y ago
> outputs 3 chars You mean codepoints or maybe grapheme clusters? Anyways yeah it’s a little more complicated but the principle of being able to truncate a string without splitting a codepoint in O(1) is still useful
55.
by rockwotj 1y ago
A friend is starting a company to do evals by just pitting models agent each other in simulations. Their teaser video is good (and humorous!) https://kradle.ai/
56.
by rockwotj 1y ago
Shortwave has had all this and more for over a year and still nothing in Gmail.
57.
by rockwotj 1y ago
Yes please don’t add another component to introduce idempotency, it will likely have weird abstraction leaking behavior or just be plain broken if you don’t understand delivery guarantees. Much better to support some kind of label or metada
58.
by rockwotj 1y ago
Came here to say these same things exactly. Best write up I know on this subject: https://use-the-index-luke.com/sql/partial-results/fetch-nex...
59.
by rockwotj 1y ago
Maybe needs a neutral party comparison :) The delivery guarantees section alone doesn’t make me trust it. You can do at least once or at most once with kafka. Exactly once is mostly a lie, it depends on the downstream system: unless going b
60.
by rockwotj 1y ago
Fun fact: this is how github adds syntax highlighting in its search bar. A while back I built the same syntax highlighting in Shortwave (an email client) using the same “view on top of a transparent input” trick. Pair it all with the follow

More ›