On 07.01.2014 01:53, Jens Alfke wrote:
> 
> On Jan 6, 2014, at 4:16 PM, Lars <[email protected]> wrote:
> 
>> It might depend on your definition of "added document". If `PUT
>> /db/docid` and `POST /db` are the only ways to "add", then do what Bob
>> said and stop at the first non-design document with a seq number of 1.
> 
> That’s still going to return documents by order modified, not created.
> 
> I think you mean “generation number”, not “seq number”, as only the first 
> document ever added to the db will have a sequence number of 1. And you’ll 
> only see a generation number of 1 if the doc’s never been updated after the 
> initial add, so this isn’t useful in the general case.

Yep sorry, I mean the number part of the revision. It being 1 is the
only way to tell addition and update apart, though, isn't it?

> 
>> It starts getting tricky if you also include in your definition
>> documents that are created by replication, and I'm not sure how to
>> handle that.
> 
> It’s not really any different. They appear in the _changes feed just as local 
> changes do.

But you only get the leaf revisions of the documents, not neccessarily
the first revision, which "added" the document.

> 
> —Jens
> 

Attachment: 0x7E86809F.asc
Description: application/pgp-keys

Attachment: signature.asc
Description: OpenPGP digital signature

Reply via email to