I am also not a fan of hyperliteral case-by-case error checking (it reminds me of my early C code), but it's A Style, one that the Golang team enthusiastically adopts, and it's unlikely to go anywhere.
More than just the normally used: if err := foo(); err != nil; { return err }
Large code Go code basis are so much better (more maintainable) for "hyperliteral case-by-case error checking". My impression that only people who havn't written large Go projects have this complaint.
Also, there is no reason you could not define a function like:
func foobar() (fatalErr, err error) object NonFatalError {
def unapply(err: Throwable) = err match {
case _: TimeoutException => Some(err)
case _: IOException => Some(err)
case _ => None
}
}
def executeWithRetries[T](maxTries: Int)(callback: => T): T =
try {
callback
}
catch {
case NonFatalError(error) if maxTries > 0 =>
logger.warn(error)
executeWithRetries(maxTries - 1)(callback)
case other: Throwable =>
throw other
}
def funcMayErr(): Int = throw new TimeoutException
val value = executeWithRetries(5) {
funcMayErr()
}
Try(executeWithRetries(5)(funcMayErr)).map(x => x + 1).getOrElse(default)
* those casually glancing over to figure what the code does * those actually trying to figure out what a given expression does, either because they are reviewing or debugging * compilers are people too
Conciseness is often overvalued and pursued to the extreme where effort is made first by the author to seek for the perfect oneliner, and then for the reader to actually check that this code is doing what expected.
Composition is important, but I don't think I found great real world examples of composition which wasn't either working only because of a tightly controlled code base or because it was just an example to prove a point.
Don't get me wrong, I love scala/haskell, I find playing with those constructs interesting and beautiful.
It's just that Go is a different thing, is a modern approach of getting back to basics, a minimal toolset for do just programming, more or less translation of thought into instructions.
And it's works very well; it's very easy to get things done quickly and the produced code tends to be easily maintainable. It's easy to have control over the memory footprint. The tooling is very mature (http://blog.golang.org/race-detector, gofmt formatting+refactoring)
That it can abstract some other "common patterns" doesn't solve this.
can easily be implemented in Go: http://play.golang.org/p/kMNqfY7LYX
What I was showing was that:
try {
foo()
} catch (ex1 ExType1) {
...
} catch (ex2 ExType2) {
...
} catch (ex ExBaseType) {
...
}
Underpowered error handling (and I'm not advocating for exceptions, persay), lack of generics (and the resultant copypasta party and interface{} runtime casting [aka, the return of void *]) are real warts in an otherwise fine language.
And I'm not just theorizing: I spend my days writing a large, nontrivial system in go.
I've used Haskell a lot before, and I'm not asking for no nulls or real ADTs (though I wouldn't complain), but generics + better typed errors would really help clean things up.
Meanwhile, a lot of us are just waiting for rust...
It just depends on what you're trying to accomplish, but most use-cases can be accomplished without exceptions. The other use-cases often indicate bad design.
As for generics, you can get 90% of the way there with interfaces. The use of `interface{}`--while sometimes necessary--is often an indicator of bad design.
In large code bases, you often don't need (or care) to know what underlying type something is. For example, you shouldn't care whether an `io.Reader` is a TCP socket, file or completely in-memory ala `io.Pipe()`.
There are times when type assertions are the best/only way to get something done, and that's why they're there, but those cases should be relatively infrequent.
Generics would make some things easier (Rust's implementation is quite nice), but it's not significantly impacting my productivity, and I certainly wouldn't consider switching languages just because Go lacks them.
EDIT: Added info about generics
type URLError {
url String
}
func (e URLError) Error() String {
return fmt.Sprintf("Invalid URL: %s", e.url)
}
This is a widely used idiom in Go.
They are used for things like out of bounds indexes, where in C it would simply be a segfault. A panic is a way of gracefully exiting a program that would have segfaulted otherwise. Correct code should check for out of bound indexes either way.
Of course this "pattern" should only be used when you're sure that that one failing goroutine won't have a cascading impact on other goroutines that are still running.
Though you raise an interesting point: Should you check array indexes if the runtime is also checking it for you?
In Java, the runtime is guaranteed to throw an exception and it is relatively rare that you would pre-check the array indexes (you might use assertions in debug code).
Incidentally, array bounds checking is actually relatively expensive, to the point where most JVMs (which use signed indexes) use an unsigned comparison trick to make it one comparison instead of two. So it does matter...
Except you don't. I haven't used recover in any of my code for a long time (more than a year). Most of the time you don't need to worry about handling panics, but you can if you really need to.
> Should you check array indexes if the runtime is also checking it for you?
In Go the generated code does it. You shouldn't do it yourself.
We're seriously drifting off track here, but if we should rely on Go to check array indexes, that seems like you _would_ want a recover block, so that we can map it to a Go-preferred error code?
There's also no such thing as a "recover block" (you're thinking of a "finally block" or "catch block", neither of which exist in Go).
If we thought you should use recover any time there might be an array out of bounds panic, we'd have designed the whole language differently. Panics should happen when things go badly wrong, and most of the time that means your program should crash.
You should use recover only in two rare cases: 1. where you're specifically using panic/recover as a kind of setjmp/longjmp (as it is used within encoding/json, for example), and 2. where you don't want a programming error to bring down your entire program, such as in the base net/http handler (although I think it's debatable whether we should have done it there; but it's done now and we can't change it).
It amazes me that there has been so much discussion over this incredibly minor and seldom-used feature. Just return and check errors (and just panic when things go really wrong) and get on with your life.
But correct code must assume that any function you call might throw; which is why you should use defer blocks e.g. to release resources, instead of C-style "cleanup at the bottom of the function". (Defer is also prettier IMHO)
The execution is suspended, the stack unwound until the first handler, the handler has access to a value that is "thrown". Stack information is preserved in order to print meaningful stack traces.
C++/java/python have syntax sugar that performs a pattern match on the thrown object to decide whether to handle it or bubble it up, while in Go you do it manually, but other that that I don't see much of a difference in the mechanics of them to justify being so pedantic about the naming of the action.
As opposed to an error, which can and will happen.
Why do people keep ignoring me when I say this? I guess the idea that the Go team are a bunch of generics-hating curmudgeons is more compelling than the reality.
However, I'm taking into account you said that they do :)
I don't blame you for not writing it up, I don't think I would either ;)
My concern is not nomenclature, but that correct code must handle paniceptions. And that it must also handle return value errors. Go code IMHO ends up spending a lot of code on error handling, I think because of this double taxation.
No, correct code doesn't have to handle panics. Panics are for programming mistakes.
At least in my world, every time I type "go", I must ensure that I am starting a goroutine that has some sensible top-level recover mechanism, and, honestly, for any Go program that actually plans on using concurrency, I think there's no alternative. You MUST handle panics. Why? Because an important aspect of Go's concurrency is maintaining composition of independent processes, and there are few things more uncompositional as a completely unrelated computation in a completely unrelated thread that trashes your entire OS process.
Panics may be for programming mistakes, but for any non-trivial code, you have some. Hopefully you can work out a better way of handling them than completely bringing the entire program down.
I am willing to assert that my code maintains enough isolation that continuing on is a reasonable thing to do. (It's a port of Erlang code anyhow. This is not a very freaky claim about such code.)
Without defer, to be correct you would have to explicitly 'recover' (and re-panic?)
You seem really hung-up on this point. Can you link to some code that illustrates your concerns?
It's even used in the standard library: http://golang.org/src/pkg/text/template/exec.go#L93
(I use the pattern myself in certain situations. It's extremely useful.)
I have exactly two panics in my ~15k line server. Both are in initialization code that will probably never get called, so it will fail very early on in the code. The rest of my code looks like this:
func getRecord(args...) (err error) {
if err := doSomethingRisky(); err != nil {
return err
}
if err := doSomethingElseRisky(); err != nil {
return err
}
... other code ...
return nil
}
func processRecord() error {
if err := getRecord(args...); err != nil {
return err
}
... do other stuff ...
return nil
}
I'm not talking about panicing instead of returning errors. I'm talking about using an idiom---which is used in the Go standard library (see my link up-thread)---to make error handling more terse when you're working with code that is otherwise profligate with checking errors.
At no point is a panic exposed to the user of a program or to the client of a library. At no point are errors ignored. The panics are kept within package and converted to error values.
Without seeing specific code I can't say for sure, but it's very unlikely that any database interaction code is best modeled with panic/recover for error handling. I'm very curious to see the source, at this point.
Using panic/recover doesn't automatically make your code nonidiomatic.
> (That the stdlib uses panic/recover in a few specific places does not make it broadly idiomatic.)
That the stdlib uses panic/recover in several places is a good indicator that "never use panic/recover" is bad advice. Note that while I agree that just because something is in the stdlib doesn't mean it's idiomatic, I also cite that this particular approach is used to make the structure and organization of code more clear. Since it's used in several packages, I claim that this is a strong hint that panic/recover is appropriate in limited scenarios.
> Without seeing specific code I can't say for sure, but it's very unlikely that any database interaction code is best modeled with panic/recover for error handling. I'm very curious to see the source, at this point.
It's really not that hard to imagine. For example: http://play.golang.org/p/fhpRLd8EHY
We seem to have some wires crossed. Let's be clear, shall we?
* The panic/recover idiom is rarely used, but it is an idiom.
* There are trade-offs involved with using panic/recover. In my sample linked in this comment, many of the functions in database/sql need to be stubbed out so that they panic. However, the cost of this is relatively small, since it can mostly be isolated in a package.
* The idiom is most frequently seen in parsing because there are a lot of error cases to handle and the response to each error is typically the same.
* While parsing is the common scenario, I claim it is not the only one. I cite that database work is profligate with error checking, and depending on your application, there's a reasonable chance that the response to each error is going to be the same. When doing a lot of it, it can pay off to use the panic/recover idiom with similar benefits as for doing it with parsing.
* There may well be other scenarios where such handling is appropriate, but I have not come across them.
I've done an unusual amount of work with parsers and have done some database work, so I've had the opportunity to bump up against the panic/recover idiom a bit more than normal. As with anything else, it can be abused. But I find it extraordinarily useful in certain situations.
* generic functions (e.g. append)
* generic data types (e.g. slices)
* exceptions (like illustrated above)
* special case syntax
* Custom event loops
* precompiler macros (very bad to use, horrible, blah blah ... except of course for the people imposing this restriction, and YES they're using it amongst other things to workaround the lack of generics in C)
...
This attitude was common in middle-90s "generic" programming languages like Ocaml, Modula-2 and others. You should simply look at Go as one of those languages and treat it as such.
If this attitude bothers you, you should look at C++0x and D.
The canonical examples are closing a file and releasing a mutex. Both have code samples here: http://blog.golang.org/defer-panic-and-recover
I'm confused as to what you guys are saying: are you saying that you don't need to handle exceptions (whether using defer or recover), or that it's better to use defer over recover? I take exception to the former, totally agree with the latter.
> are you saying that you don't need to handle exceptions
> (whether using defer or recover)
If you do still think I have gaps in my knowledge, I humbly suggest that you briefly fill in those gaps with facts; it should save you time in the long run and will likely win you a few converts!
Defer is primarily used to make sure that clean-up happens in functions that have multiple exit points (return statements). It's a convenient side effect that defers are executed while a panic unwinds the stack, but it is rarely the first thing on the mind of the Go programmer when they type "defer".
Embrace error values. Return them! Check them! Panic when shit goes really bad. That's it. If you're writing Go code and you're thinking about "throw" "catch" or "finally", you're doing it wrong. Go's features do not map cleanly to those concepts, because Go doesn't have exceptions.
This won't unlock the mutex on panic, which is observable if someone is trying to recover():
func F() {
mutex.Lock()
... do something here that panics ...
mutex.Unlock()
}
func F() {
mutex.Lock()
defer mutex.Unlock()
... do something here that panics ...
}
You should probably be using defer() all the time anyway, unless you have a good reason not to. It also helps with code evolution, in the cases where some yahoo adds a new return statement in the middle of a function.
try {
... do something that throws ...
} catch (...) {
... deferred code
}
... other code
I think it's much easier to audit this:
mutex.Lock()
defer mutex.Unlock()
mutex.Lock()
try {
... code that panics
} finally {
mutex.Unlock()
}
Also, it's trivial to make a `panic` that is unrecoverable: `go panic("broke your code, lol!!")`. This just cements the idea that `panic` is semantically different than exceptions, and should be treated as such.
You can do that by creating another function.
> Also, it's trivial to make a `panic` that is unrecoverable: `go panic("broke your code, lol!!")`. This just cements the idea that `panic` is semantically different than exceptions.
That's not different. In, say, Java, you can set the default uncaught exception handler to get the same behavior and then you can write:
new Thread() { throw new RuntimeException("..."); }You can't emulate the behavior of continuing the current block after an exception is caught. You have to recover() and copy/extract into a function any code that you'd want to run in the recover.
For example:
try {
... code that throws
} catch {
}
... other code
defer func() {
if err := recover(); err != nil {
... other code (duplicated from below)
}
}()
... code that panics
... other code
Panics and exceptions are two very different things, which is why there are different idioms in place to make working with them safe.
Right. It's a pretty trivial transformation, and that's why it's not inaccurate to call panic/recover equivalent to exceptions: you can straightforwardly express every exception-based pattern using defer/panic/recover, and also the other way round. Sometimes you have to make more functions to make panic/recover work, but that's part of the "tied in with function declarations" nature of panic/recover/defer—there's nothing semantically that deep about it because the transformation is still quite simple.
Anyway, I assume you're the same pcwalton from Rust? I really like the design of error handling so far, especially the bit about trapping conditions. From what I read, it looks failure just kills the task, instead of the entire program (like it does in go if unhandled).
I'm really looking forward to 1.0. Keep up the good work!