Hey guys, I'm Leo - one of the founders of NStack.
We started work on NStack around a year ago, to help data-scientists productionise their models and compose them together with various data sources.
We built NStack using a mixture of low-level Linux technology, including systemd-containers, btrfs, dbus IPC, along with high-level concepts such as building a DSL, type-checker, and runtime. As a functional programming company, NStack is built using Haskell and are really excited to launch our CLI and server -- along with a sandboxed demo PaaS which you can try out now.
Composition is a simple technique to build functions out of multiple simpler functions. It's heavily used in Functional programming but we think it's really useful for everyone, particularly data workers who might be gluing together lots of different modules/libraries/systems. We also use types to help check that functions can compose - this can catch schema mismatch errors early rather than waiting until you run your data pipeline to realise that the output of one function isn't _quite_ what your second function expects.
Readit News
We started work on NStack around a year ago, to help data-scientists productionise their models and compose them together with various data sources.
We built NStack using a mixture of low-level Linux technology, including systemd-containers, btrfs, dbus IPC, along with high-level concepts such as building a DSL, type-checker, and runtime. As a functional programming company, NStack is built using Haskell and are really excited to launch our CLI and server -- along with a sandboxed demo PaaS which you can try out now.
We'd love to hear any feedback.
EDIT: Sorry, it's in the subfolders.
Composition is a simple technique to build functions out of multiple simpler functions. It's heavily used in Functional programming but we think it's really useful for everyone, particularly data workers who might be gluing together lots of different modules/libraries/systems. We also use types to help check that functions can compose - this can catch schema mismatch errors early rather than waiting until you run your data pipeline to realise that the output of one function isn't _quite_ what your second function expects.
Wikipedia as some more detailed info on function composition: https://en.wikipedia.org/wiki/Function_composition_(computer...
kudos.