Add support for Possum files in S3
Currently, Possum writes hashes and S3 URIs for built functions in ~/.possum/<name> locally. To support using Possum with build tools such as AWS CodePipeline where there may not be a persistent filesystem, have the location of the .possum directory be configured as an S3 bucket.
@tomislacker Looking for your input on this as well as we improve Possum for CI/CD use.
That seems legit to me. May want to add some functionality to clear out older hashes too. I realized today my bucket was already up to ~950MB in just the couple of days I've been working on my project. Not that it breaks the bank, but could be a nuisance.
On the same token, that is my deployment bucket. For a cache bucket, it probably doesn't matter as much as the onus should be on the administrator to have an appropriate lifecycle policy setup.
I'm open for changing the behavior of writing the files to the deployment bucket. We need to consider the changes for opting between uploading and saving locally in #12 .