I’m an AI skeptic who still uses LLMs daily for development work. Here’s the workflow I’ve developed to get real value from these tools without falling into the trap of letting them do my thinking for me. Introduction Finding the right models to work with Planning with models Ensuring higher quality output Conclusion Introduction Everyone has their own preferences for integrating LLMs into their…
Getting data out of your Postgres database and into a different data store, like BigQuery or Redshift, is easy with logical replication. This guide shows you how. Overview of replication Setting up a publication Creating a replication user Wrapping things up Overview of replication Replication is the process of copying data from one database to another. There are different types of replication:…
When applications reach a certain scale, the database quickly becomes a non-negotiable consideration. Strange issues start to appear when storing millions of records and processing thousands of transactions per second. This post offers a few quick pointers for addressing issues as they arise. See what’s running Identify blocking processes Stop a process See what’s running Query the…
While a rake task or background job works fine for most data imports, they become prohibitively slow when dealing with millions of rows. PostgreSQL’s \copy command can handle these large datasets orders of magnitude faster than application-level approaches. Saving query results to CSV Importing the CSV file Diagnosing slow imports Large datasets to import Foreign key constraints Note: the snippets…
Iterate faster by using vim, tmux, and vim-slime to write and debug your code in a Read, Eval, Print Loop (REPL). Motivation Optimizing the REPL Setting up vim-slime Putting it all together Motivation Using a REPL can help you quickly iterate, but the default behavior of Ruby’s REPL isn’t exactly the best developer experience. Loading project files, reloading project files… Writing, rewriting,…
I read the book Shape Up by Ryan Singer, taking extensive notes about the methodology and assessing its viability for adoption by product teams. Bottom-line up front Pros Cons Shaping 1. Set boundaries 2. Rough out the elements 3. Address risks and rabbit holes 4. Write the pitch Betting Building Bottom-line up front Shape Up is an interesting product methodology that works well for 37signals, but…
Discover how polymorphic relationships in ActiveRecord can elegantly separate billing accounts from collaborative groups - a practical guide based on real-world implementation. Introduction Polymorphism Implementation Conclusion Introduction I was recently implementing a feature to support billing multi-user accounts when I realized my initial approach was missing something. It became clear as I…