Interesting benchmark on array computing in the browser: https://jott.live/markdown/wasm_vector_addition

One of the most interesting talks I watched in a long time. About the imminent financial collapse, the nonsense of how the software industry works, how both are connected, and what as a software professional (in particular a manager) you can do about it.

Be ready for 1h20m of bad news. 😉

https://m.youtube.com/watch?v=oWSEb0_2hb0

I can relate 😂😂😂

Interesting article about databases vs dataframes. A lot of work is being done in this area besides Modin (the authors of the article). For example, Ibis, polars or datafusion.

https://ponder.io/pandas-vs-sql-food-court-michelin-style-restaurant/

Looks like I'll be the release manager for pandas 1.5. Expect it in around a month

I had pending reinstalling my system for couple of years, but I finally had the time. Those are the main updates:

Fedora -> EndeavourOS (Arch)
Mate -> KDE Plasma
Bash -> zsh
Vim -> nvim

Each of the new software seems like a very good improvement compared to the previous.

The release candidate for pandas 1.5.0 is now ready to try, and we plan to release the final version in around two weeks. For more detailed info check the official announcement in the @pandas_dev channel.

Wise words from Carl Sagan, in the current times of scientific dogmatism

I started to write an Apache Arrow cookbook (in Rust). A first version (to be expanded) about array creation is almost ready. Feedback welcome: https://github.com/datapythonista/arrow-cookbook/blob/master/src/create_arrays.rs

Rust is showing impressive results in its way to the Linux kernel. Not long ago Linus anticipated Rust could land as a kernel language next year, or in next release. Today at Linux Plumbers conference results on a new NVMe driver coded in Rust were presented, and performance is on par with the existing C driver. But with the readability and memory safety of Rust. For context, 25 years ago there were attempts to introduce C++ in the kernel, but the efforts were cancelled not long after.

Meta (Facebook) just announced that PyTorch is moving to a new independent PyTorch Foundation under the umbrella of the Linux Foundation.

Congrats to the Pytorch team on what seems like an important milestone for the future of the project.

Today is release day for pandas. The pandas 1.5.0 packages for the different architectures are being built right now, and should be available via pip and mamba/conda in the next few hours.

While waiting for the official announcement, some interesting things about the new pandas:

- Support to the Apache ORC file format has been added 🧌
- A new function from_dummies, reverse to get_dummies has been implemented
- The documentation for pandas 1.5.0 will also be available in dark mode 🌙
- The new version will implement the dataframe exchange protocol. This will allow libraries like scikit-learn or matplotlib to receive dataframes as parameters, without caring if the dataframe is a pandas, dask, vaex, polars... object.
- Around 270 people contributed to the new release, 68% of them being first time contributors
- I was the release manager for this version 🤓

The official announcement will be made in @pandas_dev once the packages are ready