Performance and resource consumption of the PIM API implementation is relevant in several ways:
- writing to disk must be minimized (because of flash write cycle limitations)
- response times for queries, in particular caller ID lookups, must be small
enough to not be noticable by a human user; the goal is < 30ms for high
numbers of contacts (10000).
Compared to other embedded devices like phones or tablets, an IVI head unit typically has a faster CPU and more RAM (4GB?!).
The implementation tries to achieve these goals by not building up complex data structures on disk for on-disk searching. Instead it assembles the unified address book in memory and searches there.
If necessary, additional shortcuts might be implemented:
- pre-start daemon to load contacts into memory during head unit startup
- do on-disk search for caller ID while still assembling the in-memory
Before doing that, performance measurements will be needed.
Writing contact data is known to be slow, see bug #55925 and upstream https://bugzilla.gnome.org/show_bug.cgi?id=652659
Probably this doesn't matter, so not measuring actual performance and not tracking the issue separately.
-- GitLab Migration Automatic Message --
This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.
You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/SyncEvolution/syncevolution/issues/123.