[sync] Duplicates and slow sync

Chuck Hagenbuch chuck at horde.org
Mon May 29 10:01:31 PDT 2006


Quoting Jan Schneider <jan at horde.org>:

> While trying to track down why slow sync or refresh from server didn't
> work, I found some code in Backend.php that checks whether we have
> already seen events from the server on the client. The comments is:
>
> // For slow sync (ts = 0), do not add data for which
> // we have a locid again. This is a heuristic to
> // avoid duplication of entries.
>
> I understand the rationale behind it, we try to avoid duplicate events
> on the client. But doesn't this prohibit the purpose of slow syncs?
> It's supposed to *force* reloading of the device/client. Of course
> this makes development harder because we have to delete all client
> events before testing a slow sync, but the behaviour is clearly wrong
> for any other usage.
> Or did I miss any valid reasons here?

It's definitely wrong for refresh from server; I don't know for slow  
sync but your analysis sounds right.

-chuck

-- 
"we are plastered to the windshield of the bus that is time." - Chris


More information about the sync mailing list