- From: <karol.szczepanski@gmail.com>
- Date: Fri, 13 Feb 2015 10:00:59 +0100
- To: <public-hydra@w3.org>
- Message-ID: <003101d0476b$965526a0$c2ff73e0$@gmail.com>
Markus, Tomasz I feel both of you are right. Indeed I feel that we don't need a very special way of describing pagination to client. I think Hydra has already what is needed to do this. >From my experience, in most cases pagination is a call to same resources with extra parameters, defining either page+page size or offset+limit. From that point of view, I think templated links are good enough to address that. With two extra predicates with special meaning for page and page size notion (offset and limit can be easily calculated from each other) client will be informed how to craft the url to call a page of a collection. This pattern is already used for 'freetextQuery', where it has a special meaning not really bound to any resource property. Client doesn't need to know exact number of possible pages - in worst case scenario it will receive either an empty page or some other non 2XX response. Still, if needed this number can be provided with the response of the i.e. first page (which is usually called at the beginning). This approach won't need any special treating of collections and pages, and server will define whether the collection is actually pageable by providing a IriTemplate and templated links. What do you think about this? Regards Karol
Received on Friday, 13 February 2015 09:01:30 UTC