I'm looking for a clear statement of what we should expect to be saved from a data source defined in desktop when published on the data server.
For example (in beta 9), I have a very small summary table created by some upstream processes and saved to mySQL. Building the data source in desktop and then publishing to server, I see:
These changes are retained on data server:
* renaming measures and dimensions with more user friendly names (using "#", "%" in names instead of spelling out)
* changed columns that came in as alpha dimensions to numeric measures
* added calculated measure
These changes were not retained:
* manual sort of levels in dimensions to what makes sense for end user (ie maintaining order of ordered categorical variables, like manual bins)
* manual assignment of colors to dimension levels (ie we like to color our product language consistently across all reports)
The last two are examples of a class of data source properties which will help enable consistent reporting conventions across the organization.