rozenmd

I had a consulting customer refuse to believe me that his cheap housing was actually costing him more in downtime, one week to launch, i'm still working on it 4.5 years later.

4 mins critical path, 1h30m wall clock time (bunch of optional tests)

was around 40 mins on the critical path at the start of the year. gains came mostly from sharding tests, replacing webpack with vite, eslint/prettier with oxlint/oxfmt, caching node modules

@dang this title needs updating: "OpenCode Console hacked: User credentials leaked"

The opencode CLI has no impact.

You ever notice how Apple never really lets the user say "no" to their nudges?

It's always "maybe later" and never "no seriously, I don't want this, go away"

Anyway here's how Cloudflare orchestrates AI reviews at scale: https://blog.cloudflare.com/ai-code-review/

It's a managed uptime monitoring service (with support for basic HTTP checks, cron checks, and playwright-based browser checks running across AWS regions) that wires into highly available status pages.

You might want to spend time figuring out how to keep your status page online when your hosting provider goes down, thankfully for the business, others don't.

There may be many others like it, but this one is built differently: https://maxrozen.com/lessons-from-my-third-year-running-a-sa...

Still OnlineOrNot (https://onlineornot.com/), 5.5 years on.

Most recently I shipped an MCP server (w/ OAuth sign-in) recently to expose the API a bit better to agents, a TypeScript SDK and rewrote the CLI to use it. Also been reworking the terraform provider to cover the whole OpenAPI spec's surface area.

Just before LLMs became a thing I took the time to rewrite from having a private GraphQL server + public REST API to just using the REST API to build the product, super glad I did.

It was enough to just write your own scripts instead of using freely available ones to avoid detection for an extremely long time.

Same deal with Opal in Sydney - the old school opal card processes a second or so faster than a card/apple pay/google pay that you can just keep walking. Smashed into the barrier the one time I forgot my card.

> Your bank is not selling that information

Oh it very much is, (at least in Australia the major 4 banks have agreements with data brokers)

I owe my career on the Internet to my dad thinking he could block it from me when I was 11 years old.

September 26, 2025: https://blog.cloudflare.com/code-mode/

I liked Kenton's take on this: https://x.com/KentonVarda/status/2084990137180590572?s=20

Text from tweet:

Today we are releasing Cloudflare OS, a chatbot with connectors, just like every other tech company is doing.

Except actually, it's different. This is a remake of Sandstorm[.]io, my startup from 10 years ago, except this time built on Cloudflare Workers (the platform I've spent the last 9 years building) and deeply leveraging AI. This is more or less the culmination of my secret 10-year master plan.

This is a full-on personal app vibe coding platform, in which the sandbox is so secure that you can pretty much go wild -- the AI cannot introduce a significant security bug. We believe a company's security team can feel comfortable giving non-technical users permission to vibe code and then sleep soundly at night.

How is that possible? It's the Sandstorm security model, revisited. A "Gadget" is the same thing as a Sandstorm "Grain": a fine-grained app instance. For example, if you have a document editor app, each document runs as a separate instance of the app, in a separate sandbox (one "Gadget").

This means two things, both of which I think are Big Deals: 1. The platform can manage all access control, by controlling who can access the Gadget at all. There is no way the Gadget can accidentally leak itself to an attacker -- even an attacker who has access to other Gadgets based on the same app. 2. Since everyone is running their own copy of the code, everyone can freely modify their copy of the code.

Think about #2 a bit more.

What if, when you wanted a new feature in the software you are using, you could just prompt your agent to add it?

This doesn't work in the cloud Software-as-a-Service model, because you are not running your own copy of the app.

Sandstorm tried to change that 10 years ago, but the world wasn't ready, because not enough people had the skills or patience to actually modify their software. AI has changed that. Now you just ask the agent -- the same agent that you are using to help you interact with the Gadget can also modify the code of the Gadget.

And it is so fun.

shops you actually want to work at don't have "mandatory conditions", they have criteria that match an ideal candidate.

if you think you can do the job, track down the hiring manager and convince them.

The job has never been about "writing code". You're there to solve problems for the business.

I'd reach out to people you've worked with before and ask if they need help.

Writing unique, useful content that only you could have written.

Cloudflare free is the business model: https://blog.cloudflare.com/cloudflares-commitment-to-free/

Not _quite_ in this case, since the free plan improves everyone's experience: https://blog.cloudflare.com/cloudflares-commitment-to-free/

I used to keep all of OnlineOrNot's timeseries data entirely in a hot postgres db with the rest of the relational data.

Used to take a few seconds to get a week's uptime data and do some useful analysis.

Since moving to Clickhouse I think I can grab a full year's data in around 200ms (probably less if I try optimising it). Still completely blows my mind everyday.

Something doesn't add up between the pricing and the core pinning for me (or maybe I misunderstood):

Say you're using a m8i.large (2 vCPUs, $0.043/hr on spot pricing). Is the article saying you dedicate a whole physical core to one browser or am I totally misunderstanding here? If so, are you taking a loss on every browser hour?

Surprisingly rare occurrence: https://stackoverflow.onlineornot.com/

I worked on a project that used yarn from the early days all the way up to v3, it's slow as hell, but it works. They also have the supply chain protections.

Eventually we snapped and migrated to pnpm. Installs (both in CI and on local dev machines) are significantly faster. Turned out to be about a day's work to migrate with an LLM's help.

Common knowledge: https://en.wikipedia.org/wiki/Lock-up_period

the best part is Downdetector is inaccurate as hell - if AWS is genuinely down, folks get curious and search other providers, causing Downdetector to mark them as down too

Charging by the hour was one of the worst things for my career (was just starting out). Never realised at the time that I was subconsciously doing a slower job and not trying to make it faster or really improve until I left.

You can already ship single executables, my product's CLI is a Node single executable application

I don't touch AI for writing.

"Googlebook, because lets face it, your parents are only watching YouTube anyway"

Do you happen to live in Perth/Australia? https://www.cloudflarestatus.com/incidents/qs4k7rpkhspd