Remove dropped ledger columns from Edit Data columns

[nofield]

Dropped ledger columns should not be exposed to the Edit Data view, because the columns are non-editable and soft-deleted (marked as dropped and hidden, and removed from all views unless the user explicitly requests them). This PR filters dropped columns out from the list returned by GetObjectMetadata so it's never exposed in Edit Data.

Testing

Manual verification. I attempted to create a unit test for this specific scenario, but it got tricky with IsDroppedLedgerColumn not being exposed in DbColumn (only SMO Column), and trying to set specific properties in a column passed to GetObjectMetadata. If you have suggestions for how to work around these issues I'd be happy to write an automated test.

Table:

Edit Data does not expose the dropped column:

[shueybubbles]

        for (int i = 0; i < smoResult.Columns.Count; i++)

I highly recommend using smoREsult.Columns.ClearAndInitialize, passing a filter to exclude the undesirable rows on the server side and also passing the names of the properties you reference.
Otherwise perf is going to be really bad for objects with many columns.

---

Refers to: src/Microsoft.SqlTools.ServiceLayer/EditData/SmoEditMetadataFactory.cs:98 in 21470b0. [](commit_id = 21470b0, deletion_comment = False)

[shueybubbles]

        for (int i = 0; i < smoResult.Columns.Count; i++)

https://github.com/microsoft/sqlmanagementobjects/wiki/Refresh-vs-ClearAndInitialize

---

In reply to: 1249850114

---

Refers to: src/Microsoft.SqlTools.ServiceLayer/EditData/SmoEditMetadataFactory.cs:98 in 21470b0. [](commit_id = 21470b0, deletion_comment = False)

[shueybubbles]

🕐

[Charles-Gagnon]

Thanks, I created a separate issue to track the perf improvement since this PR isn't changing anything there.

#1697

@smartguest FYI

[nofield]

        for (int i = 0; i < smoResult.Columns.Count; i++)

I highly recommend using smoREsult.Columns.ClearAndInitialize, passing a filter to exclude the undesirable rows on the server side and also passing the names of the properties you reference. Otherwise perf is going to be really bad for objects with many columns.

Refers to: src/Microsoft.SqlTools.ServiceLayer/EditData/SmoEditMetadataFactory.cs:98 in 21470b0. [](commit_id = 21470b0, deletion_comment = False)

I moved the dropped ledger column filtering into ClearAndInitialize as recommended, thanks @shueybubbles. @Charles-Gagnon, David was saying that the dropped ledger column filtering as it was would be a perf impact, better to handle the filtering in SMO itself.

I'm not sure if I'm now duplicating refresh calls on the columns, though, since there's both a call to ClearAndInitialize on the Columns and the Refresh() on the entire smoResult object. I looked into if the call to Refresh() could be replaced, but the 'State' property that's being checked against on line 95 is being set during that call.

[shueybubbles]

// A bug in SMO

what bug is that?

[nofield]

Yeah, I'm not sure what that's referring to.

[shueybubbles]

                    : FromSqlScript.UnwrapLiteral(smoColumn.DefaultConstraint.Text);

Add nameof(Column.DefaultConstraintName) to the property list in ClearAndInitialize, and don't try to fetch DefaultConstraint unless smoColumn.DefaultConstraintName is non-empty.

---

Refers to: src/Microsoft.SqlTools.ServiceLayer/EditData/SmoEditMetadataFactory.cs:112 in b6e5707. [](commit_id = b6e5707, deletion_comment = False)

[shueybubbles]