This is a crazy question to ask but are we going t...
# general
This is a crazy question to ask but are we going to look into supporting NextJS 12 with the
We probably might need to implement some versioning to deal with various versions of NextJS.
poor frank is going to be trapped in nextjs hell forever
lol got a question out of the blue… what do y’all think about hosting NextJS on ie. Fargate?
it costs more, but you always get the latest feature set.. don’t have to wait for SST to support
Not sure myself, I think people have been doing it like that in my previous orgs. Controversial opinion, but I wrote my own custom resource that forwards the secrets to Vercel over their HTTP API. Means I can still get preview deploys, and sync the state between Vercel and AWS. It's always going to be hard keeping up with Vercel/NextJS ecosystem. App runner could be a less complex way (avoid fargate tasks and definition).
How does ISR work with fargate
Tbh the new features don’t sound that interesting to me. I can wait a while.
@Simon Reilly Gotcha. Yeah I have to agree the Vercel experience is pretty smooth, and they are moving pretty fast.
@Yousef I was looking at other deployment options they have here, Node.js server and Docker image, and was curious on ppls thoughts of hosting it that way. I haven’t tried it, but ISR should be supported?
I agree on the Next 12 features being weaker than 10 or 11 in terms of use case.. The middleware thing might be kind of nice, but you could probably simulate that today just by editing the cloudfront config. (as in, just make a custom lambda@edge)
@Frank, the main reasons not to use the monolith Next.JS are: • Vercel itself has a disdain for it, and they do not answer or work on bug tickets specific to the monolith server, unless it affects the dev server. • All the other reasons you use serverless over a server. • There would really be no reason to use sst for this, you could just use the CDK with no disadvantage.
poor frank is going to be trapped in nextjs hell forever
This is likely true to some extent, they tend to release quickly and have a ton of backing. Most features will just work out of the box on the existing deployment, though.
@Derek Kershner, that makes a lot of sense. Got the feedback from a couple of folks about this, and yeah the currently implementation makes a lot more sense. We will stick with it.