It's already better for some things.. EC2 image builder, for example.
Best right now to switch on a case-by-case basis only.
10/14/2021, 7:56 AM
Poor. Subnets for example are marked Deprecated with the message not implemeneted. VPCs (or perhaps it was SGs) too iirc. S3 Bucket Policy too. Loads of stuff not implement not just esoteric parts
Hoping 0.2.0 is better (pushed to GH a few days ago but marked unreleased as of yesterday)
Whilst true that you can kinda of mix and match in terms of using outputs of awsnative-resource.arn/id as inputs to aws, but the big issue I see is there is no easy way to later migrate from aws-tf to awsnative - aliases etc dont cover this scenario
Problem is a lot of stuff in the AWS-tf is buggy, potentially due TF so not blaming Pulumi, and for the APIs currently supported work flawlessly in awsnative. So it's a difficult time to onboard Pulumi. AWSNative should of had the alpha label instead of marketing writing the blog imo
I guess you could later use importids to brings across all the aws-tf stuff on a completely fresh proj/stack on awsnative as long as you can get all the props identical so no diff - can see this being a challenge
Luckily at this stage we're just building dev stacks so Im hoping by time we come to do production AWSNative is in good shape - i.e. we'd destroy aws-tf/native hybrid monster, and just deploy a fresh project all on awsnative