Groupby on multiple fields using DQL

[jdgamble555 (Jonathan Gamble)]

@MichelDiz

This is a topic that I talked a lot with Daniel a few years ago. When we saw many requests that in fact are naturally part of the client layer (business logic) and not the database.

It’s nice to have different features in DB. But the real duty of a database is to store things And fight to keep fast for query data. Since in the application we only have ephemeral states and sessions. “cookies, Redux, state, Mobx”. DBs were created to store long-term information and not for data mining. For that we have specific tools. For example Python, PowerBI or OpenRefine and others.

I think it’s cool to have fancy features, but don’t expect a DB to be responsible for everything. Otherwise, nobody needs to invent anything, since DB does everything. Right?

This part of data analysis, usually someone uses some lib in python, R or something similar. But I agree on improoving DQL and I will fight for the DQL to have cool features. That make people addicted to DQL and never want anything else

So Python would definitely be middleware, not client side. I really do feel like you want to move more things on the client side than should be there sometimes.

Real database uses do much more than “store data.” You can do analytics when the core features like math equations are there.

This particular issue would be solved by plugins / adapters one day where people would write their own algorithms (hopefully in something like custom dql with all the neo4j features). This could NOT be done on the client side efficiently and is the reason why Dgraph has a GraphQL layer… to avoid the middleware all together.

Databases need queries, and great queries, to look at the data.

J