I think there's several things at work here.
1. The validation message isn't clear enough about what is going on, especially where it's going on. It's improved in R7 a bit as it now actually says 'column dependency', which would give you a clue about what to do and where to go looking, which doesn't help you very much except to know that it is better.
2. In R6,it was decided to expose the handling of primary keys in the UI in the column dependency UI.
(I've got a doc almost ready, finally got the last, and most important, of the important reviewers to review it and I just need to get some better artwork)
3. Some times, the usage of some columns in a primary key would be assigned the usage of fact rather than identifier during the loading of the schema. You seem to have one of those cases. It seems to work better in R7, which also doesn't help you very much.
The fact that this message is only appearing now suggests that the tables involved have primary keys.
In your case "Cid" has a usage of measure rather than identifier. A human could understand that the column name suggests such a usage and not a reference to a branch of the rozzers or to a Spanish hero. I am guessing that 'Cid' is used in the primary key of the table.
You have two possible actions to perform. The action you choose will depend on the nature of the column dependency you have specified, or, if the column dependency was auto-generated from the data base table's primary key, the nature of the primary key.
One action would be determined if the query item in question is, in fact, serving as an identifier. For example, if a column in your table is used in the primary key of the table, then it is a key. As such it ought to have a usage of identifier.
If that is the case, then you need to set the usage of the query item to be identifier. You do this by selecting the query item, viewing the properties, and changing the usage to identifier.
The other action would be determined if the query item was not actually a key. Some of the cases in which this can happen are if the query item which was the key to a column dependency group was removed from the group or the order of objects in the column dependency group was changed.
The action to be performed in that case would be to re-order the query items in the column dependency group so that the query item which is the key for the group is the first in the group.
If you select the query subject and invoke the context menu, there is an item for specify column dependency. Select it and examine the column dependency and verify that it correctly maps to the primary key.
One FM best practice is to verify such properties when you bring a table into a model. This is also valid for modules.
------------------------------
IAN HENDERSON
------------------------------