differential-datalog VS datalevin

Hi,

> I wonder if it lives up to the hype.

We do think so! (disclaimer: I'm a co-founder at Feldera)

To give some more background: We are co-designing/trialing feldera with several industry/enterprise partners from different domains. Our core team also built differential datalog (https://github.com/vmware/differential-datalog) in the past. And while ddlog is used quite successfully in products today, we believe the many lessons we learned with ddlog will help us to build an even better continuous analytics platform. FYI our code is open-source at https://github.com/feldera/feldera if you'd like to try it out.

Also feel free to join our community slack channel (https://www.feldera.com/slack/) if you have more questions.

The relational model (and generally working at the level of sets/collections, instead of the level of individual values/objects) actually makes it easier to have this kind of incremental computation in a consistent way, I think.

There's a bunch of work being done on making relational systems work this way. Some interesting reading:

- https://www.scattered-thoughts.net/writing/an-opinionated-ma...

- https://materialize.com/ which is built on https://timelydataflow.github.io/differential-dataflow/, which has a lot of research behind it

- Which also can be a compilation target for Datalog: https://github.com/vmware/differential-datalog

- Some prototype work on building UI systems in exactly the way you describe using a relational approach: https://riffle.systems/essays/prelude/ (and HN discussion: https://news.ycombinator.com/item?id=30530120)

(There's a lot more too -- I have a hobby interest in this space, so I have a small collection of links)

Tutorial which I didn’t see linked in the README: https://github.com/vmware/differential-datalog/blob/master/d...

This is amazing!

Have you looked at differential-datalog? It's rust-based, maintained by VMWare, and has a very rich, well-typed Datalog language. differential-datalog is in-memory only right now, but could be ideal to integrate your graph as a datastore or disk spill cache.

https://github.com/vmware/differential-datalog

Sort of related, in my mind at least, is differential dataflow, e.g. https://github.com/vmware/differential-datalog

It’s fascinating to see so many different parties converging on Datalog for reactive apps & UI.

- There are several such talks at https://www.hytradboi.com/ (happening this Friday)

- Roam Research and its clones Athens, Logseq, use Datascript / ClojureScript https://github.com/tonsky/datascript

- differential-datalog isn’t an end-to-end system, but is highly optimized for quick reactivity https://github.com/vmware/differential-datalog

- Datalog UI is a Typescript port of some of differential-datalog’s ideas https://datalogui.dev/

Rust related https://github.com/vmware/differential-datalog

Actually, that makes me wonder whether or not differential datalog falls under that umbrella, and if it could be applied in the same way Compose is.

You can also consider other durable Datalog options like datahike or datalevin which can work either as lib (SQLite style) or in a client-server setup; if you want to play with bi-temporality XTDB is a rock solid option with very good support and documentation.

There are some Clojure-ecosystems things that are pretty cool, too, that you'd probably miss going into Haskell. lacinia is an extremely cool GraphQL library, and there are a variety of interesting datalog-based datastores which are spiritual descendents of Datomic, notably xtdb (formerly crux) and datalevin. Also as noted, you can write the front-end in ClojureScript if you want to, and there are a lot of cool libraries for that as well.

There are plenty of open source Datomic Inspired databases. Check out https://github.com/juji-io/datalevin and scroll down all the way down to “Alternatives”. There was even the beginning of a rust one by Mozilla: https://github.com/mozilla/mentat

The benchmarks in question have several implementation issues, I reported them on GitHub.

https://github.com/juji-io/datalevin/issues/created_by/caval...

Highly recommend you give https://github.com/juji-io/datalevin a chance. You can use it both as a key-value and/or relational datalog store (like datomic) but it’s very simple to set up and blazing fast!

Unlike some other commenters, I agree that graph models are usually a better fit for most data than relational models. There's been some interesting work in recent years developing this idea: in the Clojure world there's Datomic, XTDB, and a host of competitors, all of which build on work from Semantic Web/SPARQL/triplestores and logic programming. Some are even intended to be used as primary datastores: they support some amount of schema and constraints, have well-defined consistency and ACID guarantees, etc. This makes them unlike graph databases like Neo4J and others, which fill an architectural role more like Elasticsearch as a read-optimization tool. Here's an interesting talk making a case for triple-based databases.