Des de fa unes setmanes he entrat en el pou sense fons de la sobirania tecnològica en el camp de la IA, i he estat configurant el meu ordinador per a poder executar coses en local sense que les meves dades i pensaments hagin de ser enviades a algun servidor
This entry contains some scattered thoughts that I don't want to loose, so I leave it here in case they are useful too. Modeling your system In distributed systems, the devil is everywhere so it's better if we start thinking about. Sources of pitfalls: * Time * Latency
![Leptos: rust full stack [Code + Slides + Video]](/content/images/size/w600/2025/12/leptos-talk.png)
rust
A couple weeks ago I gave a talk to the BCNRust meetup group about Leptos, a web framework I have several posts about. Everything there was actually a demo powered by this library, with interactive pages, real-time polls and public collaboration. Additionally, Grzegorz explains some Leptos optomizations to power his
leptos
An update on how to configure leptos 0.8 tailwindcss 4 and daisyui 5 with trunk for csr or cargo watch for ssr
rust
Those last months I decided to learn this new fancy and hipster language, so let me share a couple of links I would now recommend. I'm coming from mostly python, but a bit of functional experience with scala, so I'm comfortable with the typing system (and
rust
Leptos is a production-ready full stack server written in rust that brings component reactivity, server functions and rust's type safety without needing to know Javascript
data
Idempotentity is not only useful in the design of data pipelines, but also as a system-wide concern, since if one step is not idempotent, the entire quality of the results will be affected. Let's look at a practical example.
So the people that know me might remember that I have a sort of obsession with diagrams. It helps me think and evolve my thoughts. I also love how git works and the importance of having code that can see its changes across time. So it's no surprise
DBT is valuable but lacking. SQLMesh promises enhancements like unit tests, environments, airflow support, change plans, and CI/CD automation, providing a better development experience. Let's see its features.
Fibonacci sequence is commonly used to show recursion, but may not be the best algorithm to compare languages and performance, since the recursive implementation is not the optimal one and languages like python will outperform rust or C.
How to use one of the best AI/ML libraries (hugginface) with Spark? Let's try it
Some pictures taken in Scotland. Use them as you please under CC BY-NC-SA and if used in public (blog, presentation , news article...), cite me or this web/post. Can't be used to train AI/ML algorithms.
The pyspark library communicates to a spark driver. If the driver dies, we'll obtain some odd error messages that will not directly indicate the root cause.
After 4 years owning two reMarkable, it's time I write the good, bad an ugly things of it.
Some pictures taken in Scotland. Use them as you please under CC BY-NC-SA and if used in public (blog, presentation , news article...), cite me or this web/post. Can't be used to train AI/ML algorithms.
Just a quick post on an error you might find when using apache spark and Avro due to a version mismatch. Software Version Spark 3.3.2, 3.3 spark-avro 2.12:3.4.0 I've you have an older version of spark (latest as of now is
If you write code and you are bored when someone mentions PowerPoint, this asciidoctor slide template is for you. Write code and get animations for free!
I've been a long time looking for an excuse to really use asciidoctor to build presentations. Now that I have the opportunities, I've been enjoying it like a child. For my memory, and for yoy, if you want to use it, here are my tips. This
Des de fa un parell de mesos he estat interessat en les dades del transport aeri, i he descobert tot un món d'aficionats (en el bon sentit) i apassionats per les dades. Sobretot, de les dades obertes en les que cada persona hi contribueix altruísticament. En voleu ser