Ask HN: If you were making a tiny social network site, what stack would you use?
Interested in what people would decide these days for small projects, with limited number of maintainers, for front-end, back-end, database, cloud providers, etc.
20 comments
Is it going to remain tiny, with a limited number of maintainers? Are those maintainers going to stay put? Is it a hobby or a business?
I would pick something popular and unsexy; make it easy to google for advice, make it easy to hire folks, make it easy to onboard them.
A cautionary tale: my employer leaned hard into Elixir; but now, most of the people who know Elixir well have left. I am probably in the top ten, maybe top five, of the remaining engineers when it comes to Elixir knowledge, and I don't know much.
I would pick something popular and unsexy; make it easy to google for advice, make it easy to hire folks, make it easy to onboard them.
A cautionary tale: my employer leaned hard into Elixir; but now, most of the people who know Elixir well have left. I am probably in the top ten, maybe top five, of the remaining engineers when it comes to Elixir knowledge, and I don't know much.
This is great advice (And in particular points out the harrowing tale of careful choices that need to be made in a 'clean slate' business).
This is a hard balance to make. Technical choices influence the kinds of staff you would want working for you but you also need to prolong/balance the interests of said staff from moving onto the next thing. (I'd love to know how businesses do this beyond just more money and share schemes but maybe this kind of thing just needs to be built into the plan)
This is a hard balance to make. Technical choices influence the kinds of staff you would want working for you but you also need to prolong/balance the interests of said staff from moving onto the next thing. (I'd love to know how businesses do this beyond just more money and share schemes but maybe this kind of thing just needs to be built into the plan)
I'm not sure that technical choices are the #1 factor in people staying or going. (They can be - if someone arrives at their job to find badly written code in a poor choice of language with zero unit tests, that certain ain't gonna help.)
I would think that interesting problems are more likely to hold staff than interesting tools to solve those problems.
I would think that interesting problems are more likely to hold staff than interesting tools to solve those problems.
I’d use whatever I’m most proficient in. In my case, that would be Node or Go on the server, Postgres as a DB, Wasabi + BunnyCDN for cheap asset hosting, Preact for any fancy pages.
> If you were making a tiny social network site, what stack would you use?
A stack that's already proven to work, that someone else runs for me at as much little cost to me as possible while I collect the metrics I need to decide if there's a business.
In here I propose using private Facebook groups.
If I were making a POC social network site, I would already have had run some user studies for the initial seed group of users to understand why the existing social network sites arn't working for them or why they havn't already used one of the existing social network platforms to create a group.
Let's assume, I find that the initial seed group of users are a group of single mothers who don't want to remarry and are perfectly fine being friends with other women - their main problem is they don't want men to bother them.
I would then create a private Facebook group and invite these single mothers and see if this private social network obtains the engagement I thought it could reach.
A stack that's already proven to work, that someone else runs for me at as much little cost to me as possible while I collect the metrics I need to decide if there's a business.
In here I propose using private Facebook groups.
If I were making a POC social network site, I would already have had run some user studies for the initial seed group of users to understand why the existing social network sites arn't working for them or why they havn't already used one of the existing social network platforms to create a group.
Let's assume, I find that the initial seed group of users are a group of single mothers who don't want to remarry and are perfectly fine being friends with other women - their main problem is they don't want men to bother them.
I would then create a private Facebook group and invite these single mothers and see if this private social network obtains the engagement I thought it could reach.
Use whatever you know best. I'd use firebase and google cloud stuff because I know how to get something working with them quickly. My personal stack would be: Firebase for storage, cloud functions with python if logic was simple and mostly client side, Flask in a VM if I thought it would need more complicated stuff (like image processing) on the server side. Bootstrap and Angular for the front end. But there's no strong reason for those other then these are proven, I know them already, there are plenty of equivalent options.
Recently I've been playing with Blazor+WASM and that's looking really promising and allows sharing C# code between client and server (and potentially a native desktop app as well), and to me is more ergonomic to program in, but I haven't used it for anything non-trivial yet.
Recently I've been playing with Blazor+WASM and that's looking really promising and allows sharing C# code between client and server (and potentially a native desktop app as well), and to me is more ergonomic to program in, but I haven't used it for anything non-trivial yet.
Like, actually tiny as in "host it yourself on a pi for a dozen friends"? That's actually something I might like to work on at some point, and I'd probably use Python,Cherrypy, SQLite, and Vue.
My other thought would be to go 100% Node.js and use Svelte, and Node might be the better idea actually, because you can choose a framework and just do exactly the project structure they want you to do, and it's pretty great.
I don't think I'd ever willingly touch PHP, the whole entire model of separate Apache/PHP/Database/etc pretty much makes anything you build depend on containers or manual setup.
I've never done anything on the cloud(I work in embedded controls) aside from a personal site on HostHatch.
My other thought would be to go 100% Node.js and use Svelte, and Node might be the better idea actually, because you can choose a framework and just do exactly the project structure they want you to do, and it's pretty great.
I don't think I'd ever willingly touch PHP, the whole entire model of separate Apache/PHP/Database/etc pretty much makes anything you build depend on containers or manual setup.
I've never done anything on the cloud(I work in embedded controls) aside from a personal site on HostHatch.
I am not a web-dev but I've always enjoyed rails.
Can someone tell me honestly if rails is still relevant?
I've been thinking about learning some front-end but am baffled by the amount of choices I need to make about the platform. I maintain a simple CRUD site that runs on PHP, I'd like to rewrite it such that when the users ask for a small change its not a multi-day adventure.
Can someone tell me honestly if rails is still relevant?
I've been thinking about learning some front-end but am baffled by the amount of choices I need to make about the platform. I maintain a simple CRUD site that runs on PHP, I'd like to rewrite it such that when the users ask for a small change its not a multi-day adventure.
Rails is still relevant and has only gotten better.
Lets say I wanted to be employable-- what would I study?
Docker, Postgres w/ proper indexes, Django. If I remember correctly Instagram just used postgres + django and may even still use it. If you can adhere to it, the 12-factor app rules may simplify the design choices in both the short and long term.
Buddypress plugin for wordpress.
add media press plugin and wordfence and go from there.
I'm hoping to see a bridge or rss export / import with mastodon and matrix (could already exist, dunno atm)
phpfox seems to still be doing things - but I haven't had time to delve into it in years.
add media press plugin and wordfence and go from there.
I'm hoping to see a bridge or rss export / import with mastodon and matrix (could already exist, dunno atm)
phpfox seems to still be doing things - but I haven't had time to delve into it in years.
I will use my most familiar stack, such as Go, Vue and MySQL. If the site really tiny, I will use some unfamiliar and popular stack.
Ruby on rails obviously. The only fully featured framework focused on “one man show” projects.
Phoenix LiveView
Postgres
Fly.io
Frontend : vue
Db :SQLite
backend： rocket
Db :SQLite
backend： rocket
not a stack, mastodon