This topic is important enough to me to start a new thread, even though previous threads have hit on it.
I am wondering if its a safe practice to do things like this:
Function 1
Open transaction
Get an Object like a line
Close transaction
Return object
End function
later on...
use object props (but do not modify object)
As I understand it, I can open an object for read-only and do whatever I want. No need to worry about a transaction. Seems to work very well so far for all kinds of objects.
If I want to modify something, I must do that within a transaction. I cannot simply change a property and have it take effect somehow. I do remember a method to "upgrade" to writeable status, but I am guessing that must be done in a transaction also.
This issue heavily affects how I structure my library of subroutines. Do I return ObjectID's or the actual object for the functions? that is the question this will answer.
I need to be sure I get started on the right foot.