
Simsek
981 posts











🎉 The wait is over🎉 The DC-ROMA RISC-V Laptop II is now ready for Fedora 41! 💡 Ready to install? Check our blog below! bit.ly/4gmfOHM Final Day of the Black Friday Sale for the DC-ROMA RISC-V Laptop II! Shop yours now bit.ly/41heqlJ #Fedora #RISCV #DCROMA





So I have this theory that @ApacheDataFusio, despite being a SQL engine, will actually enable a new breed of data systems to create non-SQL languages for working with data. Here's the idea... Before, you'd have to create not only the language, but also the planner, optimizer, execution engine, and pretty much everything. Even worse than that, you'd be tied to your new language and have no escape hatch for users that preferred SQL. Not so with DataFusion. With DataFusion, you can create the language, but get the planner, optimizer and execution engine for free. With all their complexity, performance optimizations and ongoing development effort. And you get to toss in SQL at no additional cost. So now you can iterate on user experience with your new language, while still giving everyone that wants it a fully featured SQL engine. So you're not giving up 4 decades of training and momentum, but you can experiment and innovate at the edges. If we had this available when we built the Flux language for @InfluxDB 2.0, it would have saved us years of effort. And it would have given all the users that saw Flux as a barrier to adoption a familiar option. We never had the chance to iterate on Flux developer experience and the language itself because of all the other stuff. DataFusion is going to be a game changer for innovation in the data space. More languages, more storage formats, more infrastructure projects, architectural designs, and everything else. I'm really happy that we've been able to contribute and help develop it over the last 4+ years. I think @andrewlamb1111's positioning of DataFusion being the LLVM of databases is a very apt analogy.










