Thanks. It's Rails' frontend performance stigma that made me hesitate. I suppose bolting a better FE is an option, but I won't know it until I benchmark it.
I'm a rails dev too, last year I was making a mobile web app where I was concerned about frontend, so I started making a SPA in mithril.js and using Rails as the API.
Later I abandoned the SPA and went back to rails views, using turbolinks to make it fast. In reality rails + turbolinks is fast enough for most cases, and then I can sprinkle in extra javascript or use SPA on one or two pages as needed.
The reason I think it's almost better to avoid SPA + API approach is to keep the codebase simple and easier for one person to maintain.
I’ve been going through the same thought process on a similar project in Django. Every way I look at it, the SPA route is slower and far more complicated.
I don’t know much about Rails’ perf, but by and large “SPAs are faster” only seems to hold up if you’re only measuring actions after the first page load AND your pages are very dense with heavy assets.
As others have pointed out the SPA route can introduce a huge amount of complexity. If you're a Rails dev one option worth exploring might be Phoenix + the upcoming LiveView feature.
