On Mar 23, 2010, at 7:07 AM, Kevin Jackson wrote:
> 
> The problem is that we would
> like to have urls of the form page=1, page=2 etc and at the moment we
> have page?start=[]&end=[]. 


There's a deep problem with trying to do page=N style pagination. It is 
possible but you have to do extra reduce queries to support it.

Because unless you know what the startkey is of page=N, you can't query for it. 
If you are willing to do natural keys, this is simplified, as instead of pages 
you can do Jan, Feb, March, April, etc.

The other option is to do a binary search with reduce queries, to approximately 
find the startkey of the Nth row. There's a lot of overhead here, and you can 
get weird symptoms. 

Have you ever tried browsing backwards through flickr's  recent uploads feed? 
If you aren't fast enough, people upload faster than you can browse, and you 
see the same photos over and over again. If they used linked-list or natural 
key style pagination, this wouldn't be an issue.

Chris

Reply via email to