Hey all, curious if anyone has a good workflow for SST local dev and working with s3 files. I would like to have a bunch of JSON files in a local folder, read/write to them locally when doing local dev. Then when I deploy I would like those files to be sync'd to an s3 bucket and my read/write to work against s3. Maybe everything needs to be on s3, even during local dev? Any ideas on a clean way to work this?

Typically I just use s3 locally too. All I can think of as an alternative is to abstract away the s3 access, implement a file system based adapter and then switch it based on an env var driven by the stack local property. Seems like a lot of work unless there's a really good reason not to use s3 in all scenarios. Why is it you don't want to do that?

So how do you sync the files? Like i imagine there is a folder in your SST project like

s3-stuff

how do you get that up to the bucket when you start local dev? i would be very happy to just use s3 always, just wanna make sure the workflow is there.

How do they get there in the deployed state?

That's another great question haha. My first thought was to use a script, but SST scripts run in a lambda not locally so I don't think I can get access to the files. I could write my own script and run it after deploy though. Something simple that just uses a simple sync like this https://www.npmjs.com/package/s3-sync-client Another option I was thinking is hacking it together with a StaticSite. Seeing that it already takes local files and puts them into a s3 bucket, I would just ditch the cloudfront part of it and essentially try to make it a static file s3 uploader. This wouldn't do a sync though, which isn't ideal Neither are great, which I think is the root of my problem 🙂 I would like to use s3 JSON as a data source instead of a DB. The changes are not frequent and there would be a lot of benefits for what I am doing from a simple cached file. I am just not sure the best way to get the files up there, during deploy and local dev, using SST/CDK concepts. It seems like I am probably looking for some way to run a script after local dev/deploy that has access to the CDK outputs. Although if I could get a similar

hook

syntax for

Bucket

that we have for

Function

that would be good, or even better a Bucket construct that could take a file path and sync it on deploy and

start

itself.

@Ross Gerbasi You can use CDK’s BucketDeployment construct https://docs.aws.amazon.com/cdk/api/v2/docs/aws-cdk-lib.aws_s3_deployment.BucketDeployment.html It does the S3 syncing similar to StaticSite, without the CF stuff.

@Frank does SST provide any helpers to get the root path. so when I specify a folder for the deploy like this

sources: [s3deploy.Source.asset(path.join(__dirname, 'my-website'))],

_dirname

will be somewhere in the build folder, is there an easy way to get the root of the project?