Description
I'm not sure who the correct team is to assign this to but I do feel that we need to make a recommendation on this topic.
Currently there are two schools of thinking on how to defining the type of a document with in Couchbase Server:
- Key prefix
- Type field in the document
Our examples in blogs and documentation use a mix of both. This leaves our users to make the choice which later down the line can cause them pain as only some features we have or are developing will only work with key prefixing.
- XDCR filtering
Collections(See the comment below)
It's worth pointing that N1QL is every so easier to use with a type field but can still work on Key prefix. It should also be mentioned that key prefixing from a internal view point is better as it will be cheaper than parsing the whole document.
Give the points above should we make this recommendation? This will result in having to update the documentation at the very least.