Sergio Gómez

Gotta catch 'em all!

#Pingüevo
#SalmorejoTech #SalmorejoTech2025
@aulasl.bsky.social

Hoy, 12/03/2025, se cumplen 10 años desde el fallecimiento de Terry Pratchett.

Pero como sabemos que un hombre no muere mientras se diga su nombre, ¿qué tal si decimos algunas citas favoritas de sus obras?

Empiezo yo:
"¿Qué puede esperar la cosecha, sino importarle al segador?" #GNUTerryPratchett

Thanks

Por favor, quiero un análisis comparativo. Gracias

Test GenAI-powered apps in TypeScript Evalite makes evals simple. Test your AI-powered apps with a local dev server. No API key required.

I'm building a Telegram bot that detects user intents and provides pre-programmed replies. Evalite, an AI testing framework by @mattpocock.com, has been very useful for testing intent detection and refining prompts. It's an excellent tool.

www.evalite.dev

We have a group of software crafters in my city, Cordoba, and we meet a couple of times a month. Next week, we're hosting our last meeting of the year with a mob kata at the offices of @audiense.com

www.eventbrite.com/e/kata-sesio...

Since The Fate of Atlantis, I have never seen a better game or movie about Indiana Jones than The Great Circle.

I'm making a career switch from software engineer to plumber. Got a leaky toilet? Give me a call.

ollama-ai-provider Vercel AI Provider for running LLMs locally using Ollama. Latest version: 1.1.0, last published: 4 minutes ago. Start using ollama-ai-provider in your project by running `npm i ollama-ai-provider`. Th...

Released the version 1.1.0 of the Ollama provider por @sdk.vercel.ai with support for structured outputs.

Requires Ollama 0.5.0

www.npmjs.com/package/olla...

But even though I might not solve all of them, I am clear that I won’t use ChatGPT, not unless it’s a last resort.

Programming the Advent of Code in Rust requires more time than in other languages like Python or TypeScript, and I probably won’t have enough free time to dedicate to solving the next days, especially when Christmas is closer.

Yes, I take less time to solve them, but I feel that I am learning less than in 2022, and it feels boring.

I chose to do the Advent in Rust as a way to learn the language, but after this short experience, I think that using ChatGPT to learn a programming language is not a good idea in general.

I am not solving the problems with ChatGPT, I am just asking how to do something that I could search for in the Rust API documentation.

I noticed that the experience programming in Rust was less pleasant using ChatGPT than reading the documentation and solving the problems on my own.

I completed the first 6 days of Advent of Code in Rust. The last time I tried to do it in Rust was in 2022.

This time, I am using ChatGPT as a replacement for the documentation to search for the methods I need to solve the problems.

I just completed "Historian Hysteria" - Day 1 - Advent of Code 2024 #AdventOfCode adventofcode.com/2024/day/1

Using #RustLang

github.com/sgomez/adven...

```typescript import type { NextAuthFirewallConfig } from '@aulasoftwarelibre/next-auth-firewall' const authConfig = { accessControl: [ { path: '^/(signout|settings)', roles: 'IS_AUTHENTICATED', }, { path: '^/', roles: 'PUBLIC_ACCESS', }, ], // ... } as NextAuthFirewallConfig export default authConfig ```

Updated this PoC of a firewall for @authjs.dev to support @nextjs.org 15.

Inspired by the @symfony.com firewall. Users have roles, which can be used to allow access based on matching rules or HTTP methods. Additional matching rules, such as IP, will be added.

github.com/aulasoftware...

Systems Thinking Radar - Vol. 4 Se un termostato, no un termómetro. Reducción de la complejidad. Symmathesis.

Nueva entrega del Systems Thinking Radar.
Volume 4: Se un termostato, no un termómetro. Reducción de la complejidad. Symmathesis. alfredoartiles.substack.com/p/systems-th...

Si vives en Córdoba y quieres echar un rato hablando de todo un poco, tenemos en una hora un Open Space organizado por Córdoba Crafters

www.eventbrite.com/e/open-space...

Npm capture of my ollama-ai-provider package showing over 20000 weekly downloads

Github capture of my ollama-ai-provider package showing over 850 dependents repositories and packages

That is crazy... Over 20,000 weekly downloads and used in more than 850 projects and packages.

a man wearing a white uber eats shirt is clapping his hands Alt: What color is the red horse of Santiago?

My new handle in Bluesky is @sgomez.me

Release ollama-ai-provider@1.0.0 · sgomez/ollama-ai-provider 1.0.0 Major Changes Release 1.0.0 Requires AI SDK 4.0 Removed deprecated OllamaChatSettings stop topK Removed ollama-facade

Just released the first major version of the Ollama provider for the last version of the @vercel.com SDK AI 4.0

Check it out:

github.com/sgomez/ollam...

If you need a secure proxy, you can use any ssh server you trust.

Connect to it with: `ssh -D 8080 your-server`

And configure your browser/OS to use localhost:8080 as SOCKSv5 proxy.

#til

The nth attempt to create a profile on an alternative network to "the one that cannot be named."

Here I want to avoid toxicity and not talk about politics, only about technology. In English or Spanish.