necubi

I moved my startup’s marketing site and blog from NextJS to Astro, and I’m happy with it. It’s in that middle ground—focused on primarily static sites but with the ability to still write bits of backend logic as needed.

I found it hard to get next to reliably treat static content as actually static (and thus cacheable), and it felt like a huge bundle of complexity for such a simple use case.

One downside is that you're paying for the GPU whether you're fully using it or not. It takes big queries to saturate a GH200, and if you're only using 10% of the capacity of the GPU it doesn't really matter that it's 10x faster.

In a typical company you'll have jobs, some scheduled, some ad-hoc, at a range of sizes. Most of them won't be cost-effective to run on a GPU instance, so you need a scheduling layer that estimates the size of the job and routes it to the appropriate hardware. But now what if the job is too big to run on your GPU machine? Now we either have to scale up our GPU cluster or retry it on our more flexible CPU cluster.

And this all assumes that your jobs can be transparently run across different executors from a correctness and performance standpoint.

There are niches where this makes sense (we run the same 100TB job every day and we need to speed it up), as well and large and sophisticated internal infra teams that can manage a heterogenous cluster + scheduling systems, but it's not mass-market.

You can now paste markdown directly into Google docs (Edit -> Paste From Markdown)

(I have the same workflow, via Obsidian)

I'm sorry, this comment [0] was clearly written (or rewritten) by an LLM. Sections are titled, constant m dash usages, and "Thanks again for the thoughtful look! We can zoom into any other area of your choice."

[0] https://news.ycombinator.com/item?id=46194311

The docs and the comments here are clearly LLM generated. Please don't submit AI slop to HN, or at the very least talk about it in your own words!

The commit history is legitimately insane though: https://github.com/linked-db/linked-ql/commits/master/

I feel like I'm on another planet here. Basically every new piece of software in my space (DB/query engines/streaming engines) is being written in Rust these days. There are like three production projects, total, written in Zig.

If by "exceptions" you're talking about stack unwinding (as opposed to the language-level control flow constructs like throw/catch) then Rust has always had that with panics and panic=unwind.

This doesn't have anything to do with exceptions, and the context appears to be Zig, not Rust.

The article is about how we represent errors not their control flow (i.e., exceptions).

It was the same for me growing up in the shadow of Silicon Valley in the early aughts, post dot com crash. Even when I went to college in 2008 the conventional wisdom was that there weren’t going to be any jobs in software, it was all being outsourced. I studied CS anyways, because I loved it. It was still very hard to get my first job out of college in 2012.

But then from like 2015-2022 things got crazy. Anyone with a CS degree, or even a boot camp certificate, could immediately get a 200k/year job with little effort. And people started to think this was normal, would last forever. But in fact this was a crazy situation, it absolutely could not last.

I feel for the young people who thought (or were told) that CS degrees were an automatic ticket into the upper middle class. But in reality, there’s no such thing.

Cloudflare | Rust/DB engineers | SF/AUS/SEA/NYC/LON/LIS (Hybrid) | Full-time

Hey HN, I'm hiring engineers for the new Cloudflare Data Platform (https://blog.cloudflare.com/cloudflare-data-platform). We do streaming ingest and processing, managed Iceberg catalogs, and a distributed query engine—all built on R2 and the Cloudflare Edge. We're making it ridiculously easy for Cloudflare customers to build data lakes.

I'm looking for Rust engineers with experience or interest in databases, stream processing engines, and query engines. You'll get to work on incredibly cool tech on a huge scale.

Note: we are hiring for hybrid in one of our engineering hubs (SF/AUS/SEA/NYC/LON/LIS).

If you're interested, shoot me an email (in profile) with subject HN and a brief description of your background.

Rust still supports Intel Macs just fine. The x86_64-apple-darwin target was recently downgraded to tier 2, which just means that some automated tests will no longer run as part of the standard dev cycle.

It's not particularly rare in analytical databases/query engines, nearly all of which support AsOf joins these days, including

  * Clickhouse
  * DuckDB
  * BigQuery
  * Snowflake
  * QuestDB

On macOS it’s easy—opt+shift+-.

The em-dash used to be a slightly snooty way for Mac users to announce themselves. Sad that the polarity of perception has reversed.

I’ve been typing em-dashes since I got my first MacBook in 2006 and I’m not going to let the AI companies take my beautiful punctuation away from me.

And, fundamentally it can't be opensource. Bot detection (like anti-fraud more generally) is an adversarial game that relies on hidden techniques. Open-sourcing it means you lose that advantage and make life much easier for anyone trying to get around it.

Buybacks are just a more tax-efficient way to issue dividends to shareholders (dividend issuance is a taxable event and at short-term rates, buybacks raise the stock price and those gains aren't taxable until you sell, at which point it may be long-term cap gains).

It's reasonable to be upset about the fact that this is arguably a tax dodge! But all of the other criticism of buybacks apply equally to dividends which no one seems to get upset about. Fundamentally this is the corporation saying it doesn't have a market-beating way to reinvest this capital, and it's giving the money back to its owners to more productively invest.

Cloudflare is almost entirely run on Cap'n Proto, including the entire workers platform

The Pentium 4 HT 670, released in 2005, came factory-clocked at 3.8 (https://www.techpowerup.com/cpu-specs/pentium-4-ht-670.c20)

Netburst lasted a long time as intel was floundering, before Core Duo was released in 2006.

It's actually the opposite. They're buying the assets (the software), not the company.

> To ensure continuity for users and fans, as well as to continue building what we regard as an important technology, Dark Inc has sold the assets – the Darklang language, the blog, the hosted service, the Discord, etc, darklang.com, etc – to a new company started by Dark Inc's former employees.

(from https://blog.darklang.com/goodbye-dark-inc-welcome-darklang-...)

In other words, there's a new company (Darklang Inc.) that has purchased the assets of Dark Inc. (for probably relatively little money). This clears the cap table, making it easier for the former employees (now founders) to raise money for the new corporation.

Rosetta is also very useful for running x86 Linux containers for dev workflows. Hopefully that will continue to be supported.

Fascinating article! I have many fond childhood memories of the IMAX (I guess Omnimax? Although I’ve never heard it called that) dome theater at the Tech Museum in San Jose. I probably saw “Everest” half a dozen times.

I’m also slightly embarrassed to just now learn that the opening sequence where the speakers and backing structure for the screen are shown looked so real because…it was. They weren’t projecting an image, just turning on lights so you could see back through the perforated metal screen!

Please do not post LLM comments to HN

24.04 is the most recent LTS (long term support) release; it's what users are meant to be running for anything important

As an angel, you do not get to set weird terms for your 10k check. The startup is raising on standard paper (these days almost always a SAFE) and you take them.

But no terms are going to save you from the reality that a failing startup that needs to raise more money will have to accept dilutive terms. You might be able to restrict it, but the alternative is that they can’t raise at all and shut down.

Uh… who is buying tricams and hexes these days? 100 carabiners??

The soloist had their concerto memorized (and they’ve probably performed it dozens of times before). But the members of the orchestra are responsible for playing hundreds of pages of music a week. They’ve practiced any particularly exposed or technical sections, but otherwise they’re basically sight reading for the performance.

Oxide is actually in Emeryville ;)

Aphyr didn’t test foundation himself, but the foundation team did their own Jepsen testing which they reported passing. All of this was a long time ago, before Foundation was bought by Apple and open sourced.

Now members of the original Foundation team have started Antithesis (https://antithesis.com/) to make it easier for other systems to adopt this sort of testing.

Vector is often used for observability data (in part because it's now owned by Datadog) but it's not limited to that. It's a general purpose stateless stream processing engine, and can be used for any kind of events.

(I'm the creator of Arroyo)

I haven't dug deep into this project, so take this with a grain of salt.

ArkFlow is a "stateless" stream processor, like vector or benthos (now Redpanda Connect). These are great for routing data around your infrastructure while doing simple, stateless transformations on them. They tend to be easy to run and scale, and are programmed by manually constructing the graph of operations.

Arroyo (like Flink or Rising Wave) is a "stateful" stream processor, which means it supports operations like windowed aggregations, joins, and incremental SQL view maintenance. Arroyo is programmed declaratively via SQL, which is automatically planned into a dataflow (graph) representation. The tradeoff is that state is hard to manage, and these systems are much harder to operate and scale (although we've done a lot of work with Arroyo to mitigate this!).

I wrote about the difference at length here: https://www.arroyo.dev/blog/stateful-stream-processing

Standard 4-year vesting doesn't work well for this situation. A founder leaving a worthless startup with 20% of the equity is a huge problem. The remaining founder will need that equity to offer outsize offers to senior (and eventually C-level) folks to replace them. And it's demoralizing for the remaining founder and team to be working extremely hard to make the company successful, while the departed founder reaps the rewards with no effort.

Opinions will differ here, but I think if you're leaving a pre-PMF startup you've created essentially no durable value, and should return nearly all of your equity.

I've heard of startups doing 10 year vesting for founders (with double trigger) to align this better.