Fernando Hurtado

Holy shit. It's like @fernandohur could read my mind and feel my suffering... fhur.me/posts/2024/tha…

@joaolucas_nm This is such a good point. Reminds me a bit of the old adage that first time founders focus first on product while second time founders focus first on distribution.

Reality is, helping humans at scale isn’t hard if you know what to do. Figuring out which problem to help solve, and what improves lives the most, is actually a harder problem when you have no distribution.

@kysely_ can you help me get some attention to this?

Hi @vercel apparently the kysely.dev and synthql.dev sites (both hosted on vercel) are inaccessible from Nigeria, asking around it seems you had a similar issue in the past with react.dev, could you check if something similar is happening?

@joaolucas_nm Nothing has been more effective for me than 30-45m with a barbell.

I wish I could turn my brain off sometimes, I really do

"That's Not an Abstraction, That's Just a Layer of Indirection" This resonates with me a lot. Just recently I stepped 10 layers deep to find the code that actually did something. fhur.me/posts/2024/tha…

- 2012 bug, Knight Capital loses $440M - 2017 GitLab deletes customer data - 2017 AWS S3 outage from typo - 2017 IT failure grounds 400+ British Airways flights - 2019 Google Cloud config change takes down major services - 2021 network misconfig leads to Facebook global outage

One thing about software engineering: no matter how experienced you are, there’s always a small chance of shipping something that could seriously break things.

@hadamcik @ryanels Ever heard of Figma? Photoshop? After Effects? Most video games? There are lots of products where the frontend is the main driver of value, my friend.

@ryanels If you have product where frontend is harder than backend, your product brings very little value.

Accurate 🤔

@NotionHQ Better performance

Anything goes edition: What Notion feature SHOULD we ship?

@TheWorstFounder When are you coming to Amsterdam

Next week I'll be in Tallinn, Stockholm, London and Frankfurt. DM me if you'd like to meet.

@dzhng What did you use for voice to code? Is there some kind of Cursor plugin?

Just coded a RAG pipeline in <1hr using only @cursor_ai composer, optimized via Hyde (hypothetical document embeddings) and @cohere reranker, without writing a single line of code. To make it more fun, I did the whole thing while just dictating to my computer with my voice, I basically did no typing. It's insane how far programming has come in a year.

@TheWorstFounder How do you think WunderGraph would look? What will it's product look like at that scale? and who do you think will be the customers to support that ARR?

I have found a very simple question to evaluate an early stage startup: Whatwill this company look like at 100M ARR? Which leads to great follow ups. What customers will it have? ACVs? How will it market? How will it sell? What will the product look like?

@max_sixty @prql_lang Looks like it's working again, I just got into the discord server, thx!

@prql_lang @fernandohur @fernandohur I can't get an error from the Discord link — I'm using discord.com/invite/eQcfaCm…. Could you confirm the link you're seeing that's broken?

Hi @prql_lang, your discord link at the bottom of prql-lang.org is broken, can you fix (and invite me)? What a great project btw :)

@samokhvalov @PostgreSQL What makes you so sure? Care to expand on your point of view.

the fact that pgvector is not part of the core @PostgreSQL is costing Postgres ecosystem users it's already happening and the problem is growing, I'm 100% sure

@MarcJSchmidt That's fascinating. Where can I learn more about this? ideally in the context of PG Love your work on Deepkit btw, great stuff!

@fernandohur raw SQL means here using a general purpose client driver like postgresjs/pg and write your own SQL. for an highly-optimised ORM it can communicate with the db directly in binary using JIT deserializers, which is much faster

ORMs are slow not because of some inherent law of nature, but because authors have not prioritised performance. It could be faster than raw SQL if they wanted to

@e1g @TheWorstFounder Interesting. Can you explain why?

@TheWorstFounder No, and app-level batching will almost always have higher latency.

Does application layer batching have any significant advantages over just leveraging HTTP2 for web clients?

@TheWorstFounder What are the top graphql problems? Do you happen to have a list?

The biggest problem with the common GraphQL problems is that people don't use the well known and widely accepted solutions to solve them.