For example, in Swift, if you declare a function to throw an exception, then by God every call to that function, all the way up the stack, must be adorned with a do-try block, or a try!, or a try?
I agree with him on this point. Sounds similar to how it’s in Java, and I hate it. I always wrap my exceptions in a RuntimeExceptions because of this.
I disagree with him the rest of the post. The job of the programmer is to communicate the intent of the program. Both for others and for yourself. The language provides the tools to do so. If a value is intended to be nullable, then I would like to communicate this intent. I think it’s good when a language provides this tool.
Tests don’t communicate the intent of the code. Tests can’t perfectly validate all the possible edge cases of the system either.
Checked exceptions are powerful but misunderstood. Exception types are a useful part of the facade to a module - they express to a caller how it can go wrong even if used correctly.
Runtime exceptions are typically there to express contract-breaking by callers; although as an alternative return mechanism I’ve seen them used to simplify the inner workings of some frameworks.
I think they get a bad rep because there aren’t a ton of good examples of how to use them - even the java classpath had some egregious misuse initially that helped turn people off the key ideas.
Imo, if a method require the caller to do error handling, then that should be part of the return value. For example, use optional or either. Exceptions shouldn’t be part of any expected control flow (like file not found). Exceptions is an emergency panic button.
The problem is that most languages with exceptions treat that as the idiomatic error mechanism. So checked exceptions were invented, essentially, to do exactly what you say: add the exception type to the function signature.
Having separate errors-as-return-values and unwinding-for-emergencies is a much more recent trend (and, IMO, an obviously good development).
I’m not sure why it’s “obviously” good to move from one mechanism to two: as a user I now have to categorise every path to work out which is appropriate.
What I said was less about adding to a function signature than it was about adding to a facade - that is, a system boundary, although the implementation may be the same depending on language. People typically use exceptions pretty badly - a function signature with a baggage-train of internal exceptions that might be thrown by implementation guts is another antipattern that gives the approach a bad rep. Errors have types too (or they should have), and the typical exception constructor has a wrapper capability for good reason.
My own use of jvm errors tends to follow the same kinds of patterns: I think the major fault with that model is having RuntimeException as a subclass of Exception, because it’s really intended for abandonment-style errors. (The problem is that lots of people use it instead as an exception system in order to cut down on boilerplate.)
I find it eye-opening that the author prefers callsite annotation with try (although I’m not going to argue with their experience at the time). I can see this being either “no big deal” or even “a good thing” to Rust users in particular - mutability and borrowing annotations at both callsite and definition aren’t required to make the language work afaict (your ide will instantly carp if you miss 'em out) but the increased programmer visibility is typically seen as a good thing. (Perhaps this is down to people largely reviewing PRs in a browser, I dunno.) Certainly there’s tons of good food for thought there.
That’s fine, and for that there are sum types. My own opinion differs - it’s a question of taste. Being able to bundle the handling of exceptional situations aside from the straight-line logic (or use RAIi-style cleanup) is notationally convenient.
Yes, you can do the same with monads; use the tools available to you.
The exception is part of the method signature and thus part of the return value. I don’t see a difference between using if or try-catch to validate a method call.
I think try catch often leads to messy code. It breaks the control flow in weird ways. For some reason we’ve collectively agreed on that goto is a bad idea, but we’re still using exceptions for error handling.
Say an exception is triggered. Which of these lines triggered the exception? It’s hard to tell.
We might want to handle each type of error differently. If we fail to open the file, we might want to use a backup file instead. If we fail to read the file, we might want to close it and retry with the same file again to see if it works better this time. If we fail to close the file, we might just want to raise a warning. We already got what we wanted.
One way to handle this is to wrap each line around a separate try catch. This is incredibly messy and leads to problematic scopes. Another way is to have 3 different exception types: FileOpenException, FileReadException and FileCloseException. This is also incredibly messy.
Or we can include the error in the return value. Then we can do whatever we want programmatically like any other code we write.
How is that different than what Go, for example, does? An if err != nil after each statement is just as annoying. In the end you have to validate almost all return values and the way it happens is just syntax.
I agree with him on this point. Sounds similar to how it’s in Java, and I hate it. I always wrap my exceptions in a RuntimeExceptions because of this.
I disagree with him the rest of the post. The job of the programmer is to communicate the intent of the program. Both for others and for yourself. The language provides the tools to do so. If a value is intended to be nullable, then I would like to communicate this intent. I think it’s good when a language provides this tool.
Tests don’t communicate the intent of the code. Tests can’t perfectly validate all the possible edge cases of the system either.
Checked exceptions are powerful but misunderstood. Exception types are a useful part of the facade to a module - they express to a caller how it can go wrong even if used correctly.
Runtime exceptions are typically there to express contract-breaking by callers; although as an alternative return mechanism I’ve seen them used to simplify the inner workings of some frameworks.
I think they get a bad rep because there aren’t a ton of good examples of how to use them - even the java classpath had some egregious misuse initially that helped turn people off the key ideas.
Imo, if a method require the caller to do error handling, then that should be part of the return value. For example, use optional or either. Exceptions shouldn’t be part of any expected control flow (like file not found). Exceptions is an emergency panic button.
The problem is that most languages with exceptions treat that as the idiomatic error mechanism. So checked exceptions were invented, essentially, to do exactly what you say: add the exception type to the function signature.
Having separate errors-as-return-values and unwinding-for-emergencies is a much more recent trend (and, IMO, an obviously good development).
I’m not sure why it’s “obviously” good to move from one mechanism to two: as a user I now have to categorise every path to work out which is appropriate.
What I said was less about adding to a function signature than it was about adding to a facade - that is, a system boundary, although the implementation may be the same depending on language. People typically use exceptions pretty badly - a function signature with a baggage-train of internal exceptions that might be thrown by implementation guts is another antipattern that gives the approach a bad rep. Errors have types too (or they should have), and the typical exception constructor has a wrapper capability for good reason.
I should have stressed the “opinion” part of “IMO”. What I mean is that, when I read this, it struck me immediately as being exactly correct: https://joeduffyblog.com/2016/02/07/the-error-model/
That’s a cracking article.
My own use of jvm errors tends to follow the same kinds of patterns: I think the major fault with that model is having RuntimeException as a subclass of Exception, because it’s really intended for abandonment-style errors. (The problem is that lots of people use it instead as an exception system in order to cut down on boilerplate.)
I find it eye-opening that the author prefers callsite annotation with
try(although I’m not going to argue with their experience at the time). I can see this being either “no big deal” or even “a good thing” to Rust users in particular - mutability and borrowing annotations at both callsite and definition aren’t required to make the language work afaict (your ide will instantly carp if you miss 'em out) but the increased programmer visibility is typically seen as a good thing. (Perhaps this is down to people largely reviewing PRs in a browser, I dunno.) Certainly there’s tons of good food for thought there.I’m a Rust fan, and I do think they eventually struck a pretty good “visibility vs noise” balance with
?(which was highly controversial at the time).I’m coming around to it.
That’s fine, and for that there are sum types. My own opinion differs - it’s a question of taste. Being able to bundle the handling of exceptional situations aside from the straight-line logic (or use RAIi-style cleanup) is notationally convenient.
Yes, you can do the same with monads; use the tools available to you.
The exception is part of the method signature and thus part of the return value. I don’t see a difference between using if or try-catch to validate a method call.
I think try catch often leads to messy code. It breaks the control flow in weird ways. For some reason we’ve collectively agreed on that goto is a bad idea, but we’re still using exceptions for error handling.
Consider this example:
try { File file = openFile(); String contents = file.read(); file.close(); return contents: } catch (IoException) { }Say an exception is triggered. Which of these lines triggered the exception? It’s hard to tell.
We might want to handle each type of error differently. If we fail to open the file, we might want to use a backup file instead. If we fail to read the file, we might want to close it and retry with the same file again to see if it works better this time. If we fail to close the file, we might just want to raise a warning. We already got what we wanted.
One way to handle this is to wrap each line around a separate try catch. This is incredibly messy and leads to problematic scopes. Another way is to have 3 different exception types: FileOpenException, FileReadException and FileCloseException. This is also incredibly messy.
Or we can include the error in the return value. Then we can do whatever we want programmatically like any other code we write.
How is that different than what Go, for example, does? An if err != nil after each statement is just as annoying. In the end you have to validate almost all return values and the way it happens is just syntax.