Blog concept: Sketchy optimisations

Recently a colleague was showing me a concept he was working on. He drafted a change in a fight against so-called 1+n-queries (actually for some reason unknown to me they're called n+1 queries, but my head isn't able to process the problem with just one more query after n queries…); in software development using ORMs like active record it is quite easy to make a single database request objects that when a presented within a view trigger other queries for every object because it has a relationship. Round trips to databases are generally bad as they take time.

For his change, he introduced a new class that we could seemingly reuse, with a just another (a bad code-smell) declaration of relations between objects and whether these should be preloaded when retrieving the primary object. This was in response to indeed a quite bad part of our code that entailed returning objects with counts of selected associations, but instead of counting these in the database, the current code was a…

When to use x-objects?

So I wrote a few short articles on when to use FormObjects and Jobs and ServiceObjects. The question is of course "it depends", but the leading principle I have is keep it simple. That being said, for inspiration, some suggestion for different layers to manage the application complexity from Vladimir Dementyev's talk on Railsconf:

Presentation

• Controllers (standard Rails)
• Channels (standard Rails)
• Views (standard Rails)
• Presenters
• Form Objects
• Filter Objects

Application

• Authorization Policies
• Jobs (standard Rails)
• Event listeners
• Interactors
• Deliveries
• Notifiers
• Mailers (s…

When to use Modules / Concerns?

Whenever your model gets too heavy?

The easiest way to clean up your classes might be to create smaller, more concise methods. The next easiest way of tiding up your models is moving stuff to modules (whether they are 'Concerns' or not). Modules can then be included in the final classes. It will lead to a crowded list of methods exposed on these classes, for which alternative solutions exist (Presenters, Decorators), but if you shield off private methods nicely and have a consistent way of naming things, I wouldn't be too concerned about that. Note that having many modules used in only a single class might be a code smell: perhaps you're trying to do too much with that single class.

Concerns or Modules?

When you're using Rails, you can make use of Concerns. They offer a few advantages over traditional modules, so use it whenever you're bothering recreating the same behaviour using plain old ruby Modules. I prefer consistency, so if you've adopted Concerns, use con…

When to use Form-objects?

When necessary.

It depends. By default I would advise against them; not creating Form objects to receive and validate data that could be validated by the Model directly. Even when you have a few nested attributes that belong to the main model modified, I would advise against Form objects. Keep It Simple.

But… sometimes you have more complex forms that don't fit the database-mirroring model as nicely.

When to use Job (or Worker) objects?

Always.

When you are able to do stuff async (not blocking the web-request), make it async. It will also reduce the need for a category of Service-objects. Worker or Job objects can often be called inline if desired.

Sidenote: I personally prefer the "Job" object name, a Job that needs to be performed. Worker is a name that was popularised by Sidekiq, but Sidekiq moved to Jobs as well.

When to use Service-objects?

Never.

There is of course never an absolute answer to stuff but if you are running it in a background job anyway have you considered directly writing it in a Worker or Job-object? Note that you can always run jobs async when needed.

My main objection against service objects is that all too often they are ill defined as a category. So while having fat controllers or fat models may be a bad thing, just creating a bunch of somewhat arbitrary 'Services' is not making the code more manageable.

When considering adding a 'services' directory to your app, try to think of what class of problems you want to tackle. And when in doubt, just keep messing around with the somewhat fatter models & controllers.

Introducing BrandingRepo (for Rails)

Ever had the problem that you reuse the same project for a managemable number of clients? Too few to store branding materials in a database, but more than one making it hard to keep separate branches in sync?

Introducing BrandingRepo (for Rails)

The idea is simple: create a configuration file with those files that are specific to different brands/customers and store their mods in a different repository. Repository is quite a big word here: we simply create a config/brands folder in your current branch where you can push and pull your brand specific adjustments from. All managed in the same git repository.

What it is not:

• it is not git within git.
• it is not a design system, nor has it anything to do with it (I think perhaps with a few additional hacks it can be made to work with centrally managed gems/node-modules; like here: https://twitter.com/hopsoft/status/1451358882161332225?s=10)
• it is not adding brand icons to your project

Installation

Add this …

A local .test domain for development with https using Puma-dev (on macOS or Linux)

When you maintain a few projects locally developing against localhost works good enough. npm start or rails s or python manage.py runserver or php -S 127.0.0.1:8000 will boot up a server that binds to a local port and allows you to see your work locally. The advantage of using localhost is that you don't have to bother with https-traffic as browsers don't require https for their latest features, but sometimes you need different domains to test and running multiple services distinguished by nothing more than their port numbers can become hard to manage.

To address this problem not only for websites served by the puma server, puma-dev exists. It is a spiritual successor to Sam Stephenson's Pow, which solved this problem for rack-apps. puma-dev, however, can proxy other servers as well, whether these are written in Javascript, PHP, ruby or other languages; as long as these exposes a port to 127.0.0.1, your local loopback/host you can use…

Should I use Ruby on Rails in 2021?

I'm still a big fan of Ruby on Rails. No other framework has ever made me as productive. And it is no a secret that it makes quite some other product companies very successful. Think of Shopify, Github, Basecamp, Hey, and others.

But if you'd look at at the list of most popular languages, the top 10 doesn't feature ruby anymore.

In their 2020 survey on most popular technologies, StackOverflow writes:

> Additionally, Ruby, once in the top 10 of this list as recently as 2017, has declined, being surpassed by newer, trendier technologies such as Go and Kotlin.

Also if you look at Google trends, ruby has always been negligible when compared to Python or PHP or Javascript, [the trend is downward for the ruby package manager](https://trends.google.nl/trends/explor…

Using your ruby-webmock configuration for your local test service

I recently shared an overview article about Stubbing External Services in Rails. I found it when looking for the best way to stub a pletora of services in a microservices environment. Sure, docker (or whatever) everything and run it locally / in your test suite. But unless you've plenty of disk- and memory space, this isn't always a viable option. The alternative: simulate the service. Mock or stub the endpoint.

VCR and Webmock

The go to gems are Webmock, which catches request and allows you to define the responses explicitly and VCR, which allows you to record responses, and play back.

VCR is quite cool, but as it is a recorder of earlier responses, there may be a lot of noise to dig trough when trying to make the responses a bit more generic (dealing with random token requests and what else)

For testing I pers…

Global variables in Rails

A quick note, because I was using the wrong search terms. If you want to share e.g. the current user of an app with a model you can now (since Rails 5.2) use a model inheriting from ActiveSupport::CurrentAttributes. Before you were required to pass this current user explicitly or find another way to access state.

Note that this can either be a good thing or a bad thing (tl;dr: thread-local global state makes apps unpredictable)

And even the docs warn against abusing this feature. Powerful tools can come with dangerous consequences :) Global variables are immensely powerful. Use with care. I'm not even sure if I'm going down this path…

Revisiting taming ruby's memory bloat meta-edition

There are a lot of things that I don't understand. One of these things is how memory management really works. Memory management is hard, and even though I use languages that do garbage collecting by themselves, long running ruby apps seem to run out of memory after n number of days. Even the pro’s find it quite hard. While I previously resetted the failing app every now and then, I was triggered by Mike Perham’s (creator of Sidekiq) post: “Taming Rails memory bloat”.

When you start searching for the memory bloat problem, you'll find several directions. The easiest is changing a global variable which changes the number of “arena’s” where memory allocation takes place (note: I’m in no position to explain all this, please follow the references). The fanciest, however, seems to be changing the memory allocator from glibc's default 'malloc' to jemalloc. See for example [this](https://www.speedshop.co/2017/12/04/malloc…

Prometheus for slow stats

Prometheus is a statistics collecting tool that originated from SoundCloud. Designed to be used in high performance environments, it is build to be blazingly fast. Hence, the client typically is expected to be blazingly fast as well, gathering and presenting data within nanoseconds. For Ruby on Rails applications however this has lead to an unresolved issue with the Prometheus ruby-client when the same application is forked (typical for Puma, Passenger and other popular ruby-servers). The Prometheus client collects data within its own fork before serving it to the exporter endpoint. This can or cannot be a problem. When you measuring response times, running averages from a random fork may be good enough. However, when you're also counting data over time you're having separate counters in …

Be a (unit-testing) minimalist

Still (2013) a great talk by Sandy Metz on testing, and how to do it right, without getting too theoretical. While this talk is on ruby, and it uses a Rails framework for testing, it really is applicable to any other language (only the syntax will probably be a wee bit shittier ;))

Watch Rails Conf 2013 The Magic Tricks of Testing by Sandi Metz on YouTube

(and while unit-testing is between brackets, in general, being a minimalist when writing code really is a good idea)