This week's release adds support for uploading Parquet datasets directly to Amazon S3.

When building Parquet indexers, change the outputDir option to an URL in the format s3://<bucket-name> and Apibara will automatically upload the *.parquet files to your bucket as they are created. The indexer authenticates with Amazon by using standard AWS environment variables.

Redis persistence

You can now persist the indexer's state to Redis. This is now the recommended way to persist state in production since Redis provides the same benefits as ETCD and it's much simpler to run in production.