-
The REPL: Issue 42 - January 2018
SPAs Are Just Harder, And Always Will Be
William Gross points out that Single-Page Applications add more development overhead than traditional server-rendered applications. There are more layers to code and maintain and essentially add to each browser the burden of distributed computation and data synchronization.
Things I wish ActiveRecord had after using Ecto
Ecto is a database wrapper library for Elixir. It’s design is very different that ActiveRecord – the Ruby library included with Ruby on Rails. They are often compared, because each seems to hold the majority of mindshare in it’s own ecosystem and there is a significant portion of the Ruby community interested in Elixir.
Vladimir Rosančić walks through the things he likes about Ecto that are missing from ActiveRecord. He names changesets, database constrain validation, explicit pre-loading, batch inserts, safety when loading single records and the query language itself. I find these type of comparisons really useful. They usually make clear how the choice of language or library affects the code we write.
The Modular Monolith: Rails Architecture
In this post Dan Manges details how the engineering team at Root dealt with the fabled Rails monolith and made it more modular. The achieved a healthier separation of concerns, faster builds and got rid of circular dependencies by using Rails Engines to separate the different domains in their app. They obtained a lot of the benefits often attributed to micro-services, without adding layers of network traffic (and the failure modes that come with that) in the middle of their app.
-
Structs With Keyword Arguments in Ruby 2.5
Ruby 2.5 was released a few days ago. Among the new features,
Struct
s gained the ability to be instantiated with using keyword arguments. -
The REPL: Issue 41 - December 2017
Trying to Represent a Tree Structure Using Postgres
Pat Shaughnessy writes a great 5 part series on using tree structures inside Postgres to store hierarchical data. In this first post he covers on why using a tree structure makes sense. In later parts he will cover the basics of the LTREE extension, how to install and use it and how it hooks into the Postgres internals.
Building a Distributed Log from Scratch
Brave New Geek writes the first part of a promised series on building a distributed log from scratch. in this post he focuses on storage mechanics. If you interested in why using a log is a good abstraction for distributed systems, see the referenced article The Log: What every software engineer should know about real-time data’s unifying abstraction.
Distributed systems for fun and profit
Mikito Takada writes a short e-book about distributed systems at a high level, covering scalability, availability, performance, latency and fault tolerance. The implications of different levels of abstractions, time and ordering and different modes of replication are part of the fun. Warning: After reading you might find yourself going down the rabbit hole resaearching Vector Clocks and CRDTs (convergent replicated data types). See you there.
-
The REPL: Issue 40 - November 2017
Redis Streams and the Unified Log
In this article, @brandur writes about the unified log concept and how to use Redis streams (coming soon) to build a foundation for a unified log. He covers what a unified log is good for, compares it to Kafka and provides code examples that tie everything together. This is great quality writing. I highly recommend you read the other articles on his blog. They are worth it.
CLIs are reified UIs
This articles provides some perspective about CLIs vs GUIs. The author makes a convincing argument, that CLIs make the interaction with the computer clearer, because they are more visible. This brings easier interaction because of the ability of copy, pasting, editing and so on.
Brilliant Jerks in Engineering
Brendan Gregg breaks that the implications of having a brilliant at engineering team member that is also a jerk. He breaks down the jerkiness into selfless and selfish. The post is thorough and found myself nodding along to many of the described behaviors and the problems that they cause.
-
The REPL: Issue 39 - October 2017
Floating Point Visually Explained
Sooner or later every software engineer runs into issues with floating point arithmetic precision. Fabien Sanglard explains how floating point numbers are stored and how the approximate real numbers. The post talks specifically about numbers in C, but the lesson is applicable generally.
API design: Choosing between names and identifiers in URLs
Martin Nally covers the ins and outs of using human-readable names or ids in URLs. Both have their place, even in the same systems.
10 new features in Ruby 2.5
Ruby 2.5 is expected to be released this Christmas, like it always does. Here are a few new features that will be included. There are no major changes. The language is relatively mature now and the core teams seems to be focused on performance improvements.