Working with NSCell-Based Table Views

Although most tables use NSView subclasses for cells, you might need to work with a table that instead uses NSCell subclasses. Working with an NSCell-based table is similar to working with an NSView-based table, but there are differences in how you add cells to the table and how you use Cocoa bindings.

Creating an NSCell-Based Table in Interface Builder

The steps you take to create an NSCell-based table in Interface Builder are identical to the steps you take to create an NSView-based table. For specifics, see Constructing a Table View Using Interface Builder.

After you add an NSCell-based table to your user interface, you can add individual cells either manually or using Interface Builder. To use Interface Builder to add cells:

  1. Drag the correct cell type from the object library to the appropriate column within the table view.

    Only subclasses of NSCell can be inserted into NSCell-based table views.

  2. Repeat step 1 as often as necessary to provide cells for every column.

  3. Consider setting an identifier for each column. Set the identifier in the Identity area of the Identity inspector.

    Setting a column identifier makes populating and retrieving columns substantially easier.

The Role of a Table View Data Source Object

NSCell-based table views are similar to NSView-based table views: They have a data source that adopts the NSTableViewDataSource protocol, and an optional delegate that adopts the NSTableViewDelegate Protocol to allow custome table view behavior.

To tell the table view how many rows it has to display, NSCell-based table views must implement the numberOfRowsInTableView:NSTableViewDataSource method. It must also implement the tableView:objectValueForTableColumn:row: method that returns the data displayed in each column cell. If the table view data is editable, then a method to allow the updating the model values can be implemented. Note that when using Cocoa bindings for populating an NSCell-based table view, the data source methods required for programmatic populating are unnecessary.

A table view class that conforms to the data source protocol is responsible for:

Although the NSTableViewDelegate Protocol declares all the data-providing methods to be optional, all data sources that programmatically populate NSCell-based table views must implement those methods. The methods that provide the number of rows and the content for each item within the table are optional when using Cocoa bindings.

If your app uses a data source, the data source class must adopt the NSTableViewDataSource protocol and implement some or all of the methods defined by the protocol. Even though all the methods in the data source are optional, the numberOfRowsInTableView: method is not optional if you are not using Cocoa bindings.

Providing Data to a Table View Programmatically

To populate a table view programmatically, use this process:

Populating a Sample Table

The following example populates a simple table view and enables editing. The table consists of a single column called “Names”.

The table column has had its column identifier set to name. This can be done in Interface Builder by selecting the table column, opening the Table Column Attributes inspector, and setting the identifier string. It can also be done programmatically by invoking setIdentifier: on a table column, passing the column name. The table column identifier is a great convenience when using NSCell-based table views with multiple columns (and it’s an absolute necessity when using NSView-based table views). To access a table column instance, simply ask the table view for the column by using the identifier string. In this way you eliminate the need to cache the table columns for comparison. Setting the column identifier is a simple practice that you should always perform.

The code fragment in Listing 8-1 shows the implementation of the numberOfRowsInTableView: data source method. For this simple code snippet, it’s assumed that the app’s delegate is acting as the data source object of the table view and that the sample data is a simple array of strings stored in a property declared as namesArray.

Listing 8-1  Example numberOfRowsInTableView: implementation

- (NSInteger)numberOfRowsInTableView:(NSTableView *)tableView
    // This is a defensive move
    // There are cases where the nib containing the NSTableView can be loaded before the data is populated
    // by ensuring the count value is 0 and checking to see if the namesArray is not nil, the app
    // is protecting itself agains that situation
    NSInteger count=0;
    if (self.namesArray)
        count=[self.namesArray count];
    return count;

The numberOfRowsInTableView: implementation returns the number of objects in namesArray. First it does some defensive programming. It declares a count variable and sets it to 0 and then ensures that self.namesArray is not nil. This is a good practice because sometimes a NIB containing a table view may be loaded and sent a reload message before the data itself has been initialized. Because this method returns a scalar, an exception may occur. By performing this simple test, your app can ensure that the data has been loaded and is available before any attempt is made to display the content.

Having implemented the method that returns the number of items in the table view, you must now implement the method that populates those rows. The tableView:objectValueForTableColumn:row: method returns the appropriate value to display for the requested row and column of the specified NSTableView. You can see in Listing 8-2 an example of why the table column identifier is so useful: The identity of the column that requires population is sent the identifier message and the name returned is compared to determine the data to provide.

Listing 8-2  Example tableView:objectValueForTableColumn:row: sample implementation

- (id)tableView:(NSTableView *)aTableView objectValueForTableColumn:(NSTableColumn *)aTableColumn row:(NSInteger)rowIndex
    // The return value is typed as (id) because it will return a string in most cases.
    id returnValue=nil;
    // The column identifier string is the easiest way to identify a table column.
    NSString *columnIdentifer = [aTableColumn identifier];
    // Get the name at the specified row in namesArray
    NSString *theName = [namesArray objectAtIndex:rowIndex];
    // Compare each column identifier and set the return value to
    // the Person field value appropriate for the column.
    if ([columnIdentifer isEqualToString:@"name"]) {
        returnValue = theName;
    return returnValue;

In Listing 8-2, returnValue is a local variable for storing the value returned by the method. The return value is of type NSString because all the column contents are strings. If the table had a combination of NSTextFieldCell, NSImageCell, and so on, the returnValue would typically be typed more generically, as id.

Next, the object assigns anObject to the object at the rowIndex offset in the namesArray property. Because there is a one-to-one correlation between a row and the array index for the corresponding value, accessing the object for a row is straightforward.

Finally, the returnValue variable is set to the value that’s displayed in the column. To eliminate code and simplify returning the data, it takes advantage of key-value coding.

Programmatically Editing Data in an NSCell-Based Table

For your app to edit the content of an NSCell-based table view, you implement the tableView:setObjectValue:forTableColumn:row: data source protocol method. This method is similar to tableView:objectValueForTableColumn:row:, which provides the data for the table view, but instead of requesting that you return a value for the specified row and column, it provides the new value for that row and cell.

Listing 8-3 shows an implementation of the tableView:setObjectValue:forTableColumn:row: method.

Listing 8-3  Example tableView:setObjectValue:forTableColumn:row: implementation

- (void)tableView:(NSTableView *)aTableView setObjectValue:(id)anObject forTableColumn:(NSTableColumn *)aTableColumn row:(NSInteger)rowIndex
    // The column identifier string is the easiest way to identify a table column.
    NSString *columnIdentifer = [aTableColumn identifier];
    if ([columnIdentifer isEqualToString:@"name"]) {
        [namesArray replaceObjectAtIndex:rowIndex withObject:anObject];

The example implementation in Listing 8-3 examines the column identifier method result to ensure that it’s the “Names” table column that’s being edited. This is a good practice regardless of the number of columns in a table. After verifying the column’s identity, the code replaces the object in the namesArray with the new value.

Creating Bindings for an NSCell-Based Table View

The Cocoa bindings technique used in NSView-based tables differs significantly from that used in NSCell-based tables. In an NSCell-based table you bind the table column’s content binding to the array controller’s arrangedObjects, and then configure the column’s cell’s bindings. You never bind directly to the table view’s content binding.