UITableViewDelegate and UITableViewDatasource confusion

Question

I understood before that
Delegate pattern is used only for invoking events to delegate instance and getting controls (like size / font / etc...).

Datasource pattern is only for getting data from datasource instance (like views / title / description / etc...)

But seems it was a nice illusion, after looking to Apple's UITableViewDelegate protocol I got confused because

- (CGFloat)tableView:(UITableView *)tableView heightForRowAtIndexPath:(NSIndexPath *)indexPath;
- (CGFloat)tableView:(UITableView *)tableView heightForHeaderInSection:(NSInteger)section;
- (CGFloat)tableView:(UITableView *)tableView heightForFooterInSection:(NSInteger)section;

Are delegate methods (but I was thinking that they are UITableViewDatasource methods)

Is this a dirty code from Apple, or I'm missing something important too understand difference between datasource and delegate?

EDIT: Thanks @DBD for nice answer, here is more confusion

Here is UITableViewDelegate method that returns View for drawing

- (UIView *)tableView:(UITableView *)tableView viewForHeaderInSection:(NSInteger)section;

And also there is a configuration in UITableViewDataSource

- (NSString *)tableView:(UITableView *)tableView titleForHeaderInSection:(NSInteger)section;

And oops, we can see a method that returns a View in UITableViewDataSource

- (UITableViewCell *)tableView:(UITableView *)tableView cellForRowAtIndexPath:(NSIndexPath *)indexPath;

Here we have question why cellForRowAtIndexPath: and viewForHeaderInSection: are not in UITableViewDataSource

Answer 1

This is how I've always thought about it.

UITableViewDataSource to be primary data. What are the actual contents of the table. How many rows? What is the content of row X?

UITableViewDelegate was secondary and display data. How tall should it be, should it display in the selected state, and call backs for "hey I'm about to do something."

However I admit I see some of it as a fine line (and I don't buy some of the choices)

UITableViewDataSource has titleForHeaderInSection. UITableViewDelegate has viewForHeaderInSection.

So if it's pure "data" title, it's the data source, but if includes a display wrapper with a view, it's the delegate. But wait, cellForRowAtIndexPath is a view and that's part of the data source, so why would you put viewForHeaderInSection in the delegate? While I can barely see the distinction between as "cell" as data and "title view" as delegate, I think the confusion of splitting "title" methods into different protocols is not preferable. I'm sure many might disagree with me, but it's just my opinion.

Answer 2

I think the critical distinction here arises from what you consider "data." From your question, I think you understand "data" to mean "any return value" – that is, methods which return void are delegate methods, and methods which return non-void are data source methods (since they pass something back to the sending table view).

This can sometimes be a useful approximation, but here is inaccurate. A table view's data is the contents that it displays – the stuff in the cells, the titles of sections, etc. Any other information, including that about layout (like row height) or display (like section headers) properly belongs in the delegate, since it is not about the contents of the table – merely about how to display those contents.

The two are very often related, which is why more often than not the same UITableViewController subclass implements both the delegate and data source, but imagine: you could have one object act as the data source and vend cells, then have a different object act as the delegate and provide heights for your rows based on completely different criteria. (Imagine a table where the user can resize rows, for example. You still provide the contents of each row, but the height – the delegate's responsibility – is drawing from a very different set of information.)

Answer 3

dataSource and delegate are both protocols but they are separated into two terms so that we can better understand what the methods are designed to do.

This means:

1. The dataSource protocol defines an API that supplies the data where delegate supplies the behavior.
2. dataSource is in the model layer and the delegate is in the control layer.

I think this is the correct outlook.