versioned dataset and new un-versioned anno classes

[forestknutsen1]

We are creating a few new anno feature classes within a versioned feature dataset.

1. Copy selected anno features from existing anno classes into new anno classes (all within a feature dataset) with python
2. Un-register the new anno classes with python
3. Update the reference scale for each new anno class by hand with ArcCatalog
4. Register the new anno classes as versioned with python

We have a large number of existing versions that we cannot afford to damage. We noticed that when we un-versioned the new anno all versioned editing stops working. And the feature dataset is no longer versioned. But all the versioned feature classes within the feature data still appear to be versioned, , save the new anno classes, e.g. they still have evw's (Esri versioned views) in the database. After we version the new anno classes everything appears to be working normally again.

Is this workflow safe for our existing versions and versioned data? 

[MarlonAmaya]

@forestknutsen1 

When adding new feature classes into an existing feature dataset that is already versioned, the feature dataset is no longer read (seen) as versioned. This is by design, and you will not be able to edit any feature class inside the dataset. You will need to register the dataset as versioned once again, which will now register your new feature classes as versioned. All other feature classes that are already versioned will not be affected.

https://desktop.arcgis.com/en/arcmap/latest/manage-data/geodatabases/registering-data-as-versioned.h....

When working with feature datasets, the registering as versioned needs to take place at the dataset level.

Marlon

[forestknutsen1]

@MarlonAmaya 

Thanks for the feedback. So, in our case we are copying the data into new feature classes and then un-versioning them and then versioning. This is all at the feature class level with in the feature dataset. While one cannot do this at the feature level with ArcCatalog, arcpy does appear to support it. So, our workflow is not quite the same as what you outlined. Does our workflow sound okay. 

[MarlonAmaya]

@forestknutsen1 

When you add new feature class into a dataset, they will not be versioned, however you are indicating the need to un-versioned. I would conduct the following:

1. Create a new feature class inside feature dataset
2. Copy selected anno features from existing anno classes and paste into new feature class
3. Register the feature dataset as versioned
   
   • this will register the dataset as versioned and will take care of the new feature class that was added

I am not sure if there are any implications with your workflow, but since it cant be done on the client (pro/ArcMap) side, I am surprised ArcPy is allowing it. 

Marlon

[forestknutsen1]

Okay. The work flow our contractor has put together for this is making me a little unconfutable. I can see that the features on the other side of the relationship are not getting un-versioned when we un-versioned the copied anno classes. In the past I have seen changes to in versioning follow feature classes across relationships. For example, creating a relationship between a versioned feature class and a non-versioned feature class will version everything. It would be a disaster if version data was impacted with this process.