DevLog podswiper #1

Getting the idea

I am a avid podcast listener and have the problem that I can't keep track of all the episodes I want to listen to and when I have last checked the podcast for new episodes I am interested in. The solution: Having a tool that shows me all new episodes of all my podcasts that I can swipe to dismiss or save to listen later.

Like many other developer I love to build my own tools and as I am somewhat of a domain shopper I cound't resist to buy a stand alone domain for this project: podcastswiper.com. After setting everything up for the domain (DNS Records, Webserver, Cloudflare, ...) I realized that the domain name is too long so I also bought podswiper.com which I was later informed can also be understood in a other way but I digress.

For the time being I will build on top of the YouTube API. This gives me access to all the podcasts I have subscribed, their episodes and the metadata. And I can save the episodes I want to listen to directly in a playlist.

Setting up the infrastructure

I am currently on the phase of self hosting my own infrastructure to gain a better understanding of the underlying technologies.

To get some basic protection I set up a Cloudflare Proxy in front of my own ubuntu server. On my server I run a nginx instance as a reverse proxy for my services with some basic rate limiting settings.

The app requires a database to store the podcast data and the user data. I wanted to use a SQL database and as I have the most expertise with MySQL I set up a MySQL database on the same server which very conveniently already has a mysql server running for another project.

In some future project I will give PostgreSQL a shot to see how it compares and why everyone is defaulting to nowadays.

Setting up the app

For the front end my go to choice is Next.js. I am a big fan of React and want to get a better understanding of the Next.js app router as I have previously mostly used the pages router.

Styling wise I will use Tailwind CSS as I like the very declarative nature of it. I have not yet decided whether I will use a component library or not.

To query the database I will use Drizzle ORM. In a past project I have used Prisma and was horrified by the inefficient query building. It produced queries that ran for many seconds which if I hand rolled would have taken couple of milliseconds. With Drizzle I like the sql like syntax and think that in being closer to sql it will not produce such inefficient queries from seemingly simple statements.

In order to get access to the YouTube API I set up a Google OAuth Client set to dev mode in the Google Cloud Console. This gives me 10000 credits per day for the YouTube API which should be more than enough for the beginning.