Am Dienstag, den 08.03.2011, 09:03 +0200 schrieb Marius Vollmer: > ext Arjan van de Ven <[email protected]> writes: > > > I'm sorry that your technology did not win. Actually, I'll take that > > back. I'm not sorry. Your technology did not win this round. It won > > the last round and then NOTHING good got put into MeeGo. If you really > > want to try to sell the same thing again, sorry, too little too > > late. I have a really hard time feeling sympathy for this. > > Fair enough, but let's try to not make this a "us vs them" thing. > > Let's give Tracker a fair chance: I know you do, and I also respect your > decision that you can't ship with it as the PIM backend right now, but I > still believe the 'Tracker approach' has a lot of good things to offer. > > We just need to make it work well enough, and I am certain we can.
Well said. Thank you. Little side note at the end: I've spent two years with EDS when working on Fremantle's contact's application. Therefore I believe to have some insight into what EDS can achieve and what not. For simple sync tasks like store and forget, it is almost perfect. Hard to beat the performance it's trivial architecture there. If you'd like to do select contacts by filters things become less optimal: EDS' query language is quite limited. For fast lookup you must create indexes, which are expensive to create. Actually in Fremantle's implementation of EDS most time for saving contacts is spent for updating indexes. Have to admit to not have reliable numbers right now, but as far as I remember time for updating indexes was in the scale of 80..90%. If you now also want to aggregate data, e.g. online presence, then EDS fails flat. You cannot easily remember origin of information. Automatic unmerge of contacts almost impossible. It's query language is too limited for doing those aggregations. In Fremantle we ended with keeping __all__ contacts in memory so that we could update presence information. Not exactly a scaling solution. 80k contacts surly will kill the EDS based N900. Ciao, Mathias PS: -- Mathias Hasselmann <[email protected]> http://openismus.com/ _______________________________________________ MeeGo-dev mailing list [email protected] http://lists.meego.com/listinfo/meego-dev http://wiki.meego.com/Mailing_list_guidelines
