This probably requires some server support because when people try to sort by a certain column they might expect that the result is based on all records in the database, not just the subset currently displayed in the UI.

We won't be able to do server side sorting on any column, just on a specified subset. Here' the backghround info.

http://directory.fedoraproject.org/wiki/Simple_Paged_Results_Design

We can use a VLV for, say uid and Last Name, and allow windowing of results on that. Each (VLV) index we add slows down the addition of data, but without it , there is no way to perform both sorting and windowing on a single entity.

Investigate in September.

Design document:
http://www.freeipa.org/page/IPAv3_Directory_Browsing

My vote will be to start with option 2 and add option 3 for primary use cases.

We should probably split this ticket to first implement option 2 across and then add tickets to move to hybrid mode in particular areas.

This ticket depends on ticket #215. To be optimal, option #2 also needs to use server session. Without it option #2 will work but be limited by server's and browser's memory size.

It seems to be possible to implement basic pagination feature now. The option to retrieve primary keys has been implemented (ticket #1262). The default size limit imposed by the LDAP server seems to be big enough to handle most cases. The UI needs to specify --sizelimit=0 to bypass the IPA server size limit and instead use the LDAP server size limit.

However, if the number of entries is very large and the LDAP server size limit cannot be configured to include all entries the IPA server will still need to use paged results to get all primary keys.

master:
- df6bd6a
- 9405e1a