Remote content

[adelevie]

Dunno if this is putting any carts before the :horse:, but I would like the to separate metadata from the actual document content. E.g. something like this should be possible:

{
// ...
"data": {
    "content": {"content-type": "text", "url": "http://mys3buck.et/foo.txt"}
// ...
}

"ingestors" would have to know how to handle this and make the HTTP request.

[konklone]

I could see content being a separate field, actually, but yes, this would need to be handled elegantly.

Though I wonder if there's always just one item of content? Man, this is why specs get out of control. Forget I said that.

But remember that I said that including content inline could also be a good goal.