I'm working on a new portfolio piece to demonstrate that I can quickly learn and utilize new tech. This will provide a great way for potential employers to see how I handle wading into completely unfamiliar territory.
To do this, I'm using the Deno runtime instead of the Node.js runtime. I've never used any runtime other than Node for my backend or js first projects.
A benefit of Deno is that it supports typescript out of the box without any additional buildsteps. I'm a huge fan of this and it makes me much more likely to choose typescript as a tool for my personal projects.
In addition to the Deno runtime, the Deno foundation has released a brand new server-side rendering framework called Fresh. This framework is competing with the likes of Next.js and Sveltekit. It uses typescript and preact right out of the box.
I'm going to use these tools to create a basic social media website. My initial goal is to make a lightweight clone of Twitter, but the scope may change as I encounter difficulties or find better ways to implement features.
You can keep track of my progress using these repos:
Frontend Fresh server repo: click here
Backend REST api repo: click here
I've now written a few web apps in Go. My latest creation features Go on the backend using Gin as a lightweight web framework and gorm as my ORM. Gorm really surprised me with how easy it is to use. I still have a lot to learn about how it works, but it appears to manage migrations in a really intuitive way.
My database is PostgreSQL and it has been pretty easy to use. I just spin up a docker container with Postgres installed and connect to it with gorm.
On my front end is a Preact server that gets data from my back end through a REST api. I looked into GraphQL and it was pretty interesting, but has a decent learning curve and was not nearly as simple to implement.
For now, I'm working on authentication and exploring the ecosystem in Go. I would likely never roll my own auth in a production app, but going through the process of designing auth from scratch really helps me understand what all those third party libraries are doing anyway.
Sessions have been on the top of my list. My current goal is to make a Reddit style social media app that can scale to many users.
This is my tentative plan for now. As long as I design my front end and back end to be stateless, they should hopefully be able to handle arbitrary requests without knowledge of the previous ones. A caveat is session storage, but I can store session data in redis.
I may have to scale up the database layer too with multiple read only databases and a few write only databases, but I'm not 100% sure how to go about that yet. I'm not particularly experienced at scaling web applications to crazy amounts of users, but even this basic plan should ensure that I can use something like kubernetes and docker to scale the number of servers.
Improving
September 07, 2022
Getting better at something is a skill. I've learned that the hard way over the past few months. Deliberate practice is key to improvement and I've been doing a whole lot of deliberate practice.
Most of my time is split between brushing up on data structures and algorithms, working on portolio projects, or applying for jobs. It has been exhausting, but I am improving at an unprecedented pace.
Writing Go code has been eye opening in a lot of ways. I hardly understood concurrent design and rarely bothered with pointers (even when the language I used supported them). But Go emphasizes these things. I know the applications and cost of passing by value or reference. I understand channels and goroutines and why they are such powerful tools.
The next step is to use these concepts and cement them in my mind through a project. I've been playing with a toy blockchain implementation and that has been fun, but I really want to create a web app utilizing microservices. Something that can scale through concurrent design.
Anyway, I'm rambling a little bit. The point of this post is just to document some of my improvements as I go.
As of right now, I need to brush up on my front end skills. I plan on trying out ember.js and if that isn't a good fit I'll go back to React/Vue/Svelte. In addition to that, I want to learn FastAPI and write some more Python. Go is great, but I need to keep those Javascript and Python skills up to date.
It's pretty funny that I would write about specialization just after writing a post about wearing every hat.
But I'm not talking about that kind of specialization. I'm talking about picking a set of tools and really building my proficiency with them. As I've explored development and programming for the last 15 years or so, I've mostly done it because I was curious. It was a hobby of mine to track down the newest, shiniest things and play with them.
In this sense, I was always trying out something new. I've written pet projects in dozens of programming languages and I've used hundreds of different libraries. Unfortuantely, this method doesn't really lend itself to mastery. If I want to really get good with my tools, I need to keep using the same ones.
So — I've decided that my backend tool of choice is Go. Go is a very powerful programming language with out of the box features that I love. There's no need to download a framework to get going with Go. The standard library has most of the tools you need to create a strong backend design.
However, Go can be a little overkill in some scenarios. In those cases I want to utilize Python with Django or FastAPI. Python is a great language for productivity. I can get things up and running super fast, and I love this. In addition, Django is pretty opinionated framework. This means that it is sort of an "all tools included" approach. Everything I might need for a web app should have some native implementation in Django. If Django is too heavy weight, I have plenty of other options, too.
Finally, on a related note, I've also decided to really buckle down and master PostgreSQL. It's an incredibly powerful, open-source, DBMS and basically an industry standard.
I'm still making up my mind about my front end tools. Ember.js looks promising and server side templates are always an option. In the future, I'll likely consume an API with react or something, but that's for later. For now — mastering Go and Django.