Normally we want to set this maximum amount bigger than the amount of data that we are expecting during the session - if you set the maximum amount bigger than expected data, then there's no need to do batching. am i missing something?
Is it perhaps possible to investigate the amount of data before transfer? - you can simulate a sync and on the ChangesSelected event, you can figure out how many rows has changed down to the each table. When you get the row counts, simply throw an error
and that would cancel the sync.
in your example above, if you have a 10mb change dataset, and you want to apply changes in 1mb increments, you will have to invoke the sync 10 times.
assuming you can apply the changes partially, how do you tell your user which rows comprise the 1mb you partially applied out of the 10mb change dataset? if you have related tables, this becomes even harder. do you just bring over the parent table first?
how about their child records? 1Mb parent and zero child? 500kb parent and whatever you can fit in 500kb for the child rows?
