Tickets :: Comment on [#12770] Sync "loop" of Kolab categories <-> Horde tags sync

* Your Email Address
* Spam protection	Enter the letters below: . . ..__.._..__ \ / \|[__] \| [ __ \/ \__\|\| \|_\|_[_./
Comment	> Hi, > > consider this scenario with a new, empty calendar: > > 1. User inserts the first event. The event is tagged with "foobar". > This tag is inserted into the tagging backend. > 2. User creates another event with a foreign Kolab client. > This event is tagged "Foobar" (noticed the capital 'F') > 3. kronolith syncs in the foreign event. The Kolab sync tells > the tag backend to store the tag(s) for the new event object, > it does a case insensitive search for an existing tag label. > -> it picks the existing row id of "foobar". > > > The next time Event::synchronizeTags() runs, it notices > the tags given by the Kolab driver for the second event > differ from the tag backend ("Foobar" != "foobar") > -> A sync is triggered. > > I have a productive calendar with about 2.000 events. > The sync "loop" produces 3.6000+ INSERT / UPDATE statements > on every click. Yikes. > > Please consider the attached patch, which makes the > tag comparison case insensitive. > > If the patch gets approved, we need to fix > the tag sync of the other applications, too. > > Another option would be to make the tagger backend > case sensitive, but that would lead to duplicate entries > because of lower / uppercase spelling. Not really an option. > > Without the patch, browsing around in the calendar > takes about seven seconds for each click on a box with a SSD. > With the patch it's about one second. > > Note: I removed the strcoll() based sorting, it is of no use, > array_diff() does not care about the order of elements. > > Cheers, > Thomas >
Attachment
Watch this ticket

Saved Queries