Emblogginated: On the Effectiveness of FIXMEs presumably.de/on-the-effecti…

☎️ looking for the very best software engineers in Europe — €10,000 Referral bonus! ☎️ The last months have been pretty crazy. Being part of @ycombinator, growing to $1M ARR in just 8 months — we know to continue at this pace we need to find the very best people to help us build this company! We are looking to find 3 more people to join our in-person team in Berlin. Two Software Engineers and one Forward Deployed Engineer. Link for details in the thread

Announcing: Voice AI course and online community ... @swyx and I are hosting a month-long technical deep dive into Voice AI and Voice Agents. Our goals are to: ➡️ cover all the lessons we've learned over the last two years building realtime, conversational AI, ➡️host fun sessions with all our favorite people who have relevant knowledge to communicate, ➡️and build a long-term online community.

published beep-boop 0.1.2 - audiovisual feedback for test runs now easy to install via npm github.com/pesterhazy/beep-b…

Another React drama got me curious to check out the docs. I'm still not buying the whole "you need a framework" idea, especially since listed arguments are server side rendering, routing and data fetching. None of these has to be complex problems unless you willingly agree to bury yourself in frameworks and libraries that try to solve every potential case.

At @rocicorp we've always had a design doc tradition (inherited from Google I suppose). Since we are a distributed team, we would do design docs up front that are fairly detailed (typically 3-5 printed pages and taking maybe a week or two of work for each new major component. We'd discuss and iterate them as a team, then work proceeds independently - w/o blocking code review - and we stitch together at the end. This might sound crazy to many ears, but when you're working on databases and sync engines and things like this - that are pretty subtle - we have found it makes sense to think about things at a high level first, before code. This always worked on our prior projects. For @rocicorp_zero, we found this approach breaking down. The design doc system is kind of predicated on the idea that the people reviewing the docs have a big picture understanding of the whole system and can understand the new component in relation to that. But we didn't have this with Zero. None of us did. It was too big a project to fit in any of our heads at once early on. And we found we couldn't design (or review the design of) piece A without understanding piece B, C, and D first. We also found it was difficult to design some parts "on paper" without the help of the typechecker and the deeper understanding that comes from real code, a keyboard, and unit tests. We ended up – not really purposefully, but sort of out of desperation - writing several versions of the IVM pipeline as prototype/throwaway code. Several of us took a swing at it 😂. And through this kind of iterative/explorative process we got more of an intuitive feel for the system. Once we had this, we started working on real code (ironically w/o a design doc at all – there still isn't one for ivm). But we had to accept that some parts of the system we didn't know exactly how to do. We had to take a sort of leap of faith that we'd figure them out when we got there.

Excited to join telli (YC F24) as Founding Engineer - this video explains what we do in 60 seconds linkedin.com/posts/y-combina…

why use Classes in TypeScript? A quick gist gist.github.com/pesterhazy/c… by our very own @pesterhazy

I'll be in LA for the weekend, for the first time. What's the one thing I shouldn't miss?

Is anyone else on 🦋? I've just created an account and it's hilarious—exactly like twitter but fast and simple

finally @pesterhazy and I can use our favorite commit message w/o ppl getting mad at us

John Cutler's Wicked Loops - the best 8 minutes in project management. I think about this short video a lot

In San Francisco until the end of the month, anyone want to meet up for a coffee?

TIL that node-postgres returns `COUNT(*)` (which are BIGINT in postgres) as strings (!) by default. What a footgun!

Hetzner is offering an S3-compatible object store (currently Beta), and the pricing is insane compared to AWS. Storage: 4X cheaper — $4.9/TB-mo vs $21/TB-mo Traffic: 50X cheaper — $51/TB vs $1/TB And no request fees. AWS charges $0.4-$5/M requests. docs.hetzner.com/storage/obj…

Anyone have any experience with @temporalio? Its abstractions (workflows, activities) are very tempting

Code review is both a get-out-of-jail-free card for compliance and a get-stuck-in-a-swamp card for velocity.