4 ms·
>I don’t care about static or dynamic types,(...) I think that RMDB is the simplest and most reliable in theory and practice But... RDMBs are often statically
by flavio81 8y ago
>I don’t care about static or dynamic types,(...) I think that RMDB is the simplest and most reliable in theory and practice
But... RDMBs are often statically typed (you specify the column type when creating it), and strongly typed (they often don't do auto-conversions of data types.)
- robto 8y agoI think their point is that it's the ability to express relational algebra in a natural way that's important, and the static/dynamic types thing is orthogonal to that.
- lincpa 8y agoWith reference to the database, as long as I use spec to strictly define (standardization) the core data model, I can ensure the correctness of the system. I've turned the traditional work of ensuring code correctness from in-code type system validation to data self-validation. Turn the work into verifying the core data model instead of validating the input and output data types of the function. A system requires only a few core data models, and development is built around the core data model. Persistent data structures ensure that the modification of the immutable big data model are high performance and memory efficient. In addition, using my pure function pipeline data stream (https://github.com/linpengcheng/PurefunctionPipelineDataflow https://github.com/linpengcheng/PurefunctionPipelineDataflow) makes debugging, parallelism and extension very simple.
- lincpa 8y agoSimilar to industry, verify that all finished products meet the standards before entering the warehouse. Also similar to databases, verify their compliance before data enters the database. That is "Data as a service, Warehouse as the core, operates around it".