Tickets :: [#11064] Turba Browse is very inefficient

6.0.0-alpha12

6/12/25

Summary	Turba Browse is very inefficient
Queue	Turba
Queue Version	Git master
Type	Enhancement
State	Rejected
Priority	1. Low
Owners
Requester	slusarz (at) horde (dot) org
Created	03/09/2012 (4843 days ago)
Due
Updated	06/27/2012 (4733 days ago)
Assigned
Resolved	06/27/2012 (4733 days ago)
Milestone
Patch	No

06/27/2012 09:07:52 PM	Michael Slusarz	Comment #5 State ⇒ Rejected	Reply to this comment
Going to reject, since Jan says this is not going to be possible with current code.

04/05/2012 05:01:24 PM	Jan Schneider	Comment #4	Reply to this comment
The actual problem is that 1) name fields are usually composite fields and 2) there is the user preference that defines the name format for sorting. For the filtering to reliably work, this would require composing the fields on the database level. I doubt this is going to be possible in a portable way.

03/09/2012 08:20:31 PM	Michael Slusarz	Comment #3	Reply to this comment
[Show Quoted Text - 11 lines][Hide Quoted Text] Browsing in Turba (at least via SQL driver) will query ALL contacts on every page, regardless if you are filtering - for example, by a letter when using the alphabet pager. It is the View code that is doing the filtering. This is incorrect behavior - the query should be filtered at the DB level instead - it reduces overhead and SQL is precisely designed for this kind of filtering, so it is the proper tool for the job. Driver->search should support criteria like limit(n) and %like% or startwith('a') This probably applies to more backends. The problem, at least as I can tell, is that the way the Turba Views are designed. The Browse View is passed a search result, and it is only in that view that the filter (from URL parameter data) is determined. In other words, the filter can't be applied at the time the search is done based on the current architecture.

03/09/2012 08:09:58 PM	Ralf Lang	Comment #2	Reply to this comment
Browsing in Turba (at least via SQL driver) will query ALL contacts on every page, regardless if you are filtering - for example, by a letter when using the alphabet pager. It is the View code that is doing the filtering. This is incorrect behavior - the query should be filtered at the DB level instead - it reduces overhead and SQL is precisely designed for this kind of filtering, so it is the proper tool for the job. Driver->search should support criteria like limit(n) and %like% or startwith('a') This probably applies to more backends.

03/09/2012 08:01:23 PM	Michael Slusarz	Comment #1 Priority ⇒ 1. Low Patch ⇒ No Milestone ⇒ Queue ⇒ Turba Summary ⇒ Turba Browse is very inefficient Type ⇒ Enhancement State ⇒ New	Reply to this comment
Browsing in Turba (at least via SQL driver) will query ALL contacts on every page, regardless if you are filtering - for example, by a letter when using the alphabet pager. It is the View code that is doing the filtering. This is incorrect behavior - the query should be filtered at the DB level instead - it reduces overhead and SQL is precisely designed for this kind of filtering, so it is the proper tool for the job.