Post by @polotek@social.polotek.net

@polotek@social.polotek.net

2025-12-04 23:02 UTC

Part of the reason for this is that I believe in "deploy early and often" when building real projects. I don't like waiting until I'm done to figure out deployment. I like to see it live and what that's gonna feel like. It also allows me to show other people the work in progress. It's almost always overkill. But it's helpful for me personally because if I get bogged down with the stuff that's not fun late in the project, I'll probably give up and never get it out there.

Replies (3)

@polotek@social.polotek.net 2025-12-04 23:07

I'm really annoyed at database options right now. Keeping a local database in a container feels bad. It risks losing data. Probably not a big deal for personal projects, but it causes me anxiety. The project is immediately not "low maintenance" anymore. I like managed databases, but getting a separate PaaS to talk to RDS is not a thing. And if the PaaS you're using offers their own managed databases, that's usually the place where you hit the paid tier and the project is no longer free/cheap.

Open ##2300285
@OmegaPolice@hachyderm.io 2025-12-05 00:13

@polotek@social.polotek.net Agreed. Also, how would you end-to-end test without some form of deployment? And if you did that late, you'd miss out on the advantages of writing tests first.

Open ##2300302
@evana@hachyderm.io 2025-12-05 01:40

@polotek@social.polotek.net another reason to do this early is so that you convert the knowledge of how to run the application into code, rather than keeping it in squishy memory (where it is inevitably not documented well enough for when you come back 6 months later). So for me, doing the packaging / CI is basically a defense against future abandonment of the project by myself.

Open ##2300303