How to handle error with Realm during writing?

Question

I'm used to working with SQL database, I'm new to Realm and, so far, I'm really impressed by the ease of use of this new mobile database. But there is something that I really don't undestand: how to handle error throwing?

Take this simple example:

I want to store in a Realm DB some market stocks.
Each stock has a "symbol" as a unique identifier: APPL for Apple Inc, TSLA for Tesla Motors Inc, etc.
I believe it would make sense to declare these symbols as primary keys, since it's not possible to have multiple times the same symbol in the database
When user clicks on a symbol (among a list of symbols), this symbol is saved in db.

In this Raywenderlich tutorial, it is said: "To simplify the code required in this tutorial, you’ll be used try! when calling Realm methods that throw an error. In your own code, you should really be using try and do / catch to catch errors and handle them appropriately."

So according to the following pattern:

do {
    try realm.write {
        realm.add(symbol)
    }
}
catch let error as NSError {
    print("Something went wrong: \(error.localizedDescription)")
}

So far it makes sense.
And if a user clicks on a symbol that is already in the database, I get (very logically) an error:

*** Terminating app due to uncaught exception 'RLMException', reason: 'Can't set primary key property 'symbol' to existing value 'APPL'.'

The problem is that this error is not catched during run time: I've got a crash instead.

My question is not about how to avoid such a crash, I understand of course that it's easy to avoid it by doing a simple testing before writing anything in the database :)

My question is how do we do in Realm to catch potential writing errors?
Am I doing something wrong?

Answer 1

do/try/catch in Swift catch Swift Errors, which are an entirely distinct thing from Objective-C exceptions. Realm follows Foundation's pattern for error reporting: API misuse errors throw exceptions which are not intended to be caught (and can't be caught in Swift), and recoverable errors throw Swift errors (or use NSError out parameters in Objective-C).

Adding an object with a duplicate primary key is considered API misuse, so it's a fatal error as the route for "handling" it is to fix the bug in your code. An example of a recoverable error which would produce a Swift error that would be caught by catch is running out of disk space while trying to save the new data.

Answer 2

There are two distinct types of error: programmer error, and expected errors. Realm handles these two types of errors differently.

Programmer error covers cases such as accessing objects from incorrect threads, out-of-bounds collection access, attempting to insert an object with a duplicate primary key, and so forth. Realm responds to these programmer errors by throwing an Objective-C exception. These exceptions are not intended to be caught and handled as they're indicative of a bug in a Realm user's code.

Expected errors cover things like files not existing, permission errors, failures to write to disk. These are things that can happen even with correctly-written code due to factors outside of the programs control. These are exposed as NSErrors and can be caught from Swift using do / try / catch. Note that Realm's Objective-C API has some convenience methods that don't expose the NSError. In these cases an expected error will throw an exception as there's no other avenue for Realm to communicate the failure. Rather than attempting to catch and handle exceptions raised by these convenience methods, you should use the variant of the API that returns an NSError instead.