Please find the attached patch to fix the RM #4030 - IDENTITY column not recognised.
- Added support for IDENTITY column for PostgreSQL >= 10.0
A few issues unfortunately:
- There's an extra space in the generated SQL for an identity column on a table, right before the end comma (in both the CREATE and reverse engineered SQL.
- I cannot make an IDENTITY column a primary key through the UI, nor does it reverse-engineer that property in the SQL if I create it via SQL (it does properly set the switch value though).
This issue has already been logged earlier, but I will fix this with this patch.
- After creating an IDENTITY column, there should be a dependency on the sequence, but I don't see this listed.
If the Show System Object is enabled, then only you can see.
- We should consider the auto-created sequence a system object, and hide it in the treeview by default as it's an implementation detail.
How, can I identify those as a system object? I tried but couldn't find a way.
- If I click on an IDENTITY column in the treeview, the reverse-engineered SQL just shows the plain datatype.
Fixed.
- Can we reasonably support the sequence_options clause?
I will look into it.
To show the sequence option, we have 2 options.
We can provide a simple textbox or all the options listed with textboxes same as Sequence dialogue.
So, which one is preferable?
Much as I hate to add to the workload, this is a GUI designed to prevent the need to know the SQL syntax, so it's got to be individual listing of the options I think.