Search Index definition VS Population
When we create an index we define it with a JSON schema.
When pushing documents into the index we have to adhere to that schema exactly.
In our domain that JSON document may contain other attributes which are important but are not candidates for indexing. Currently we have to pull out the fields from our document and hand those on to Azure Search for indexing.
Latency, traffic, other implications aside it would be nice if Azure Search only looked for the indexed attributes and ignored everything else.
The mergeOrUpload document action should give you the behavior your are looking for. You only need to include the fields you care about and a new document will be created if one doesn’t already exist for the Id specified.