Streams are often underestimated and skipped as possible solutions. In many cases, we created much more complex solutions than their streams counterpart. Why?
It's hard to answer, but in this presentation, I would like to tell you a story about how we started to use FS2, without sacrificing purity and code readability. (https://github.com/functional-streams-for-scala/fs2)
4. Pump in ML data to HDFS
• Instead of keeping files on S3 (compressed JSONs)
we would like to have them on HDFS stored with
optimized format like Parquet
• We need to support full and incremental dumps from
our data ocean (Vertica)
• Simplify the current solution and provide
“predictable” memory usage.
54. Best practices
There is one exception, if your
application based on Streaming solution
(ex. Http4s) you can return Stream as
the result of operation
(there are benefits !)
55. Best practices
There is one exception, if your
application based on Streaming solution
(ex. Http4s) you can return Stream as
the result of operation
(there are benefits !)