You could implement some sort of sparse map. E.g. discard 9 out of 10 marks for anything more than 20 marks back. If they actually go back that far again, you re-request from the nearest mark with a larger row count.
And I would definitely add behavior analytic in this case. It may well be that 14 times out of 15, they do not actually need those extra marks at all. Then, you could argue that the 15th time is a special use case served differently (e.g. with export, separate replica, or whatever). Regards, Alex. ---- Sign up for my Solr resources newsletter at http://www.solr-start.com/ On 3 February 2015 at 10:19, tedsolr <tsm...@sciquest.com> wrote: > Oh, I know I have problems! My (b) option of reversing sort and using the > current cursor mark is not working. It gets off by one record. > > paging forward: > pg 1: docs 1-10 > pg 2: docs 11-20 > pg 3: docs 21-30 > > now paging backwards: > pg 2: docs 10-19 > > I'll go back to tracking all the cursor marks. > > > > -- > View this message in context: > http://lucene.472066.n3.nabble.com/Reverse-deep-paging-tp4183554p4183674.html > Sent from the Solr - User mailing list archive at Nabble.com.
