On 21/02/14 17:37, Graham Cobb wrote: > Of course, the reason is because we are abusing the username, which is a > sync property, to do the job of a source property (to specify the > account for activesyncd). Should we change the eas backend to use (for > example) databaseUser to specify the account? > > Or am I still confused (quite likely)?
Further investigation has led me to discover that the carddav/caldav backends work the same way. They use username and password (and also cannot be set when just configuring the source). So, I am obviously confused. It seems that in both the webdav and activesync cases, we are specifying a folder (which must be in a particular user's account) in the source, but we are not specifying the access control for that account in the source. Wouldn't it be more logical to either: 1) Keep database as is, and use databaseUser/databasePassword for the access control, or 2) Modify the database to only specify the folder name and put the username, password and url in the peer? The first case makes webdav and activesync feel like local backends. The second makes them feel like SyncML peers. The current situation doesn't feel like either, to me. I also note that the current behaviour prevents running a SyncML client or server with either webdav or activesync as the backend database underneath. But presumably that is not a design goal (only file, kde or evolution are allowed?). Graham _______________________________________________ SyncEvolution mailing list [email protected] https://lists.syncevolution.org/mailman/listinfo/syncevolution
