4 minutes read | 832 words by Ruben BerenguelSome links are affiliate links
Happy new year!
The plan for this year is to read/listen 50 books. Why? My book reading list, after cleanup and updating of all my reading devices (and paper books, too) is at 99 books. I want to make a significant reduction and so far the years of 30-ish books have made no difference. I think going for half of it (and trying not to add to the list!) should help.
Part of my new year project management changes involve moving from a Things-centered task management to an Obsidian-centered one. I wrote this Scriptable plugin (and another one, actually) to make this process smoother and better-looking.
This is just a “promotional blurb”, but highlights what shiny things dbt-databricks brings: use delta by default, take advantage of Databricks' Delta MERGE operation and, more importantly, use Photon to speed up queries.
I find it slightly funny that they ditch pre-built feature stores due to the high infrastructure maintenance and tuning cost they can incur (they point at Kafka and Spark), but instead rely on a custom built solution that relies on Snowflake and Snowflake streams 💸💸💸.
I have been following Tom and his art and comics since I found out he’s the author of one of my favourite pieces, Gnome research. Also worth listening to this presentation. In any case, this article is a short, typical interview with an artist. Tools, setup, etc.
I was very interested in Quantified Self several years ago (found a post in this own blog from 10 years ago about it), but the author here actually tracked a myriad metrics for 10 years. The analysis is very interesting, and the persistence should be rewarded.
Based on quite a few interviews, Liz Wiseman (also author of Multipliers) analyses what separates impact players (a term borrowed from sports performance) from excellent contributors. It highlights several traits and how to improve in them. Good stuff.
From HBR, may be 🔒 (I’m a subscriber). Managers used to be “good” when they managed and evaluated performance well. Later, managers are “good” if they train and coach well. The rise in remote work has moved the needle to managers being good at reading employees feelings and being empathetic. Or at least this is the premise here.
I may have already posted this piece by Hillel Wayne. Decision tables are probably the simplest formal method available, and also extremely handy to model strange business requirements. As long as there are not many state changes.
Fibers are used in most modern concurrent processing systems in Scala. This is an explanation of why they are surprisingly fast in the case of Cats Effect 3.0. I think the first paragraph about “stupid examples” is about another framework’s propaganda (scroll to Fast and correct).
Many suggestions, ideas and tips for using SQLite in unusual ways. Hacker News comments have sometimes very interesting information. The post they are about analyzes the trade-offs in using SQLite for “real” web applications, and why it’s a pretty good option in many instances.
A Data Contract should work as an API between your “real data” and its consumers. You can think of it as a guaranteed schema, extending all the way up to the producers. Having data contracts eases the work of any consumer since they’ll know the data is guaranteed to have certain properties (format, columns, types) that keep its code or reports working even after changes that extend but not alter the contract.