A look at UNQL
They should have tried harder
I have just read the UNQL-specification I tried to understand what UNQL offers to deal with unstructured and distributed data.
What I see, is that the syntax is almost identical to SQL and is not even using features of SQL that have been around for years. (Explicit join vs implicit join)
Basically UNQL just adds the ability to specify documents in JSON format and to access properties using the dot notation. (Navigation)
In many respects the spec is very unclear.
- Evaluation of properties
- How Joins shall be performed on collections in a distributed document database.
- Grouping of documents.
- Aggregation (Sum etc)
- How links between documents shall be handled
However, these are the bits, that make the design of a query language for unstructured (NOSQL) data really hard. So I doubt, that these details can be easily solved.
In its current state, UNQL is just a limited SQL that allows you to navigate into a document.
A more detailed explanation of my critique will follow in the next posts.