Some time past, Sparx - on a recommendation from Ripple - allowed columns to be created for database Views . These columns would be for information only and, specifically, would not (at that time) be used for DDL generation.
It would seem churlish now to complain that the functionality provided doesn't work. So, I'm not being a churl and therefore I'm not complaining, but merely making some observations that might help Sparx provide better support for View Columns in the future.
Firstly, at present, you can create these View Columns either by dragging and dropping from existing tables (my preferred method) or by directly creating them via the <Context Menu>|Attributes... menu item. When you edit them, EA brings up the Attributes dialog, rather than the Columns dialog, even though they are clearly stereotyped «column». As a side-effect of this, a Not Null Column is rendered as <ColumnName>: <ColumnType> {bag} - because of the use of the AllowDuplicates field in t_attribute to store the Not Null. The Attributes of a View need to be rendered as Columns.
For Columns in a View, the Not Null, Unique and Primary Key properties are not relevant and so should be removed on the "drop".
It might be useful for the Intial Value Property (which has no meaning in a View) to be used to store the column source - either directly from the table (as in <Table/View>.<SourceColumn>) or the derivation expression if more complicated. For a View, the Inital Value property (probably) shoudln't be rendered.
Thoughts? Votes?
Paolo[size=0]©2007 Paolo Cantoni, -Semantica-[/size]