Merge Replication Question

This is for SQL MOBILE:

We are looking into using merge replication for our project.

Right now we have a set up that whenever a user logs in it creates a publication with specific filters based on the user's configuration settings and permissions.

The publication is created and a snapshot is generated, I run the replication monitor and open up the subscription that is downloading data from the publication.

The subscription downloads data just fine and when I push a new snapshot the subscription only downloads the changes.

Here's my question:

When I made a change on the database server and did NOT push a new snapshot and the subscriber tried to download from the publication, instead of downloading nothing (since no snapshot was pushed hence the data in the snapshot is unchanged) it downloaded everything (by everything I mean everything on the snapshot).

So can someone explain to me how this is supposed to work?

To my understanding the subscriber gets data from the snapshots.

Situation:

new snapshot (10,000 rows) - Subscriber reports (10,000 inserts).

changed data and push new snapshot (30 changes) - Subscriber reports (30 inserts)

changed data and NO snapshot - Subscriber reports (10,000 inserts).

PS. I do not reinitialize a subscription.

I think your point of confusion is in your statement "the subscriber gets data from the snapshots".

The snapshot contains a moment-in-time image of tables, constraints, and data from a publication.

During replication, the subscriber is first compared against the latest snapshot to ensure he is at the

right generation of schema and at some point in the past, received the moment-in-time data the snapshot

contains. After that, any new changes on the publisher are sent to the subscriber, followed finally by

changes on the subscriber being uploaded.

So your subscription gets compared to the snapshot and then received all the changes on the server since

the last time the subscriber synchronized.

Darren

|||Thank you.