Conflict management - Synchronize datasets efficiently

[planemad]

It oftens happens to me (possibly because i'm on a not so reliable developing world net connection), that my uploads stops responding and i need to cancel it. This usually causes synchronization errors on reupload, the server having newer versions of objects than what i have.

I now have to synchronize the entire dataset, which causes a whole deal of downloading, because its downloading all areas from the start of my session, even those that have been uploaded ok. Ideally, it should only start synchronizing from the area that has the object which is out of sync.

Anyway, after waiting for the 120 area downloads to sync my data, i get an error "No data found in this area". I close the dialog, try to reupload, and the same error saying the server has a newer version. Clearly whatever area was downloaded to sync my dataset was lost because of the 'no data found' error. Now i have pretty much no way to upload my data but to sync the 450 objects individually.
upload-error-sync-upload..

Whats needed:
1) Start sync only from the area in which the problematic data exists, not from start.
2) Allow user to specify how many areas to sync. if there have been 40 area downloads, he may want to sync only 10 which could potentialy have a problem, not everything.
3) skip objects out of sync, upload the other stuff, maybe from reverse order. problematic objects should not be blocking the upload of clean data.

[Cobra]

There is an option to update only the modified objects (file -> update modified).
You can also select some objects and update this selection.

When uploading you can do the same. Select some objects and upload them. But it would be much better if you could upload more often so there won't be so many objects to upload and potentially update.