Ask HN: What’s the first thing you implement on a new project? | Hacker News Reader

Ask HN: What’s the first thing you implement on a new project?

For me, the first thing I always implement (that is always a bit custom per project) is localization because it’s the hardest thing to bolt on later. But I’m curious, what are some of the first things other people implement on a new (greenfield) project?

View on HN

CI/CD, which like I18n is harder to bolt on later. I only greenfield at home though (never get to at work, not in 20 years).

No matter what the project is, the first thing I do is what I call "flying a kite across the chasm" - meaning what ever is the purpose of the software, I create the barest implementation of "user starts working" to "user has completed their work" which is the purpose of the software. This is not a GUI or anything a user would use, it is the data transformation from input to final raw output. Once that is in place, the overall stress of "can it be done?" is answered and the rest is being smart about filling out the details. FWIW, my career has been in research, where I'm answering "can it be done?", "can it be scaled?", "can it be financed?" types of questions.

All the git stuff, .gitignore, initial commit, github remote setup, etc.

Its so much better than backing up to ZIP files on floppy disks. 8)

Depends on the project and effort required for each component. I’m working on a website right now that really serves as a user-friendly way to view specific data. Because the data is static, the first thing I did was upload it all to Firestore.

Usually, I’d use mock data and set up Firestore once I’ve achieved the desired level of functionality.

I used to create a Git repo as a first step, but stopped doing this because 1) I’m not worried about a hard drive failure during the initial development, 2) I want to get straight into making progress on the core idea, 3) I currently do all development on one laptop.

I do web stuff, and part of my discipline is getting a hello world server and write the long copy for the homepage.

Or I just scramble to get something that works just for me, to make sure it makes sense, even if it clearly won't work for others w/o tons of modification.

Authentication. It’s not always a good habbit. I recently started working on a no-code web page builder project, got bogged down in authentication, then realised I should work on the core functionality first, so I switched to that.

Edit: Before starting to code, I register a domain. It’s not a great habit because I’m now paying renewals on domains I don’t use. It builds up over the years.

For Android, testing. It's quite common to get caught on a architecture that's difficult to test later and crashes for obscure reasons.

Authentication. I work Rails btw...

configure.ac

Reply on news.ycombinator.com