Does Haskell require a garbage collector?

Question

I\'m curious as to why Haskell implementations use a GC.

I can\'t think of a case where GC would be necessary in a pure language. Is it just an optimization to reduc

Answer 1

Your intuition that this has something to do with purity has some truth to it.

Haskell is considered pure partly because side effects of functions are accounted for in the type signature. So if a function has the side effect of printing something, there must be an IO somewhere in its return type.

But there is a function that is used implicitly everywhere in Haskell and whose type signature doesn't account for, what is in some sense, a side effect. Namely the function that copies some data and gives you two versions back. Under the hood this can work either literally, by duplicating the data in memory, or 'virtually' by increasing a debt that needs to be paid back later.

It's possible to design languages with even more restrictive type systems (purely "linear" ones) that disallow the copy function. From the point of view of a programmer in such a language, Haskell looks a little impure.

In fact, Clean, a relative of Haskell, has linear (more strictly: unique) types, and that can give some idea of what it would be like to disallow copying. But Clean still allows copying for "non-unique" types.

There is lots of research in this area and if you Google enough you'll find examples of pure linear code that requires no garbage collection. You'll find all kinds of type systems that can signal to the compiler what memory might be used allowing the compiler to eliminate some of the GC.

There is a sense in which quantum algorithms are also purely linear. Every operation is reversible and so no data can be created, copied, or destroyed. (They're also linear in the usual mathematical sense.)

It's also interesting to compare with Forth (or other stack based languages) which have explicit DUP operations that make clear when duplication is happening.

Another (more abstract) way of thinking about this is to note that Haskell is built up from simply typed lambda calculus which is based on the theory of cartesian closed categories and that such categories come equipped with a diagonal function diag :: X -> (X, X). A language based on another class of category might have no such thing.

But in general, purely linear programming is too difficult to be useful, so we settle for GC.

Answer 2

Let's take a trivial example. Given this

f (x, y)

you need to allocate the pair (x, y) somewhere before calling f. When can you deallocate that pair? You have no idea. It cannot be deallocated when f returns, because f might have put the pair in a data structure (e.g, f p = [p]), so the lifetime of the pair might have to be longer than return from f. Now, say that the pair was put in a list, can whoever takes the list apart deallocate the pair? No, because the pair might be shared (e.g., let p = (x, y) in (f p, p)). So it's really difficult to tell when the pair can be deallocated.

The same holds for almost all allocations in Haskell. That said, it's possible have an analysis (region analysis) that gives an upper bound on the lifetime. This works reasonably well in strict languages, but less so in lazy languages (lazy languages tend to do a lot more mutation than strict languages in the implementation).

So I'd like to turn the question around. Why do you think Haskell does not need GC. How would you suggest memory allocation to be done?

Answer 3

GC is "must have" in pure FP languages. Why? Operations alloc and free are impure! And the second reason is, that immutable recursive data structures needs GC for existence because backlinking creates abstruse and unmaintainable structures for human mind. Of course, backlinking is blessing, because copying of structures which uses it is very cheap.

Anyway, If you don't believe me, just try to implement FP language and you will see that I'am right.

EDIT: I forgot. Laziness is HELL without GC. Don't believe me? Just try it without GC in, for example, C++. You will see ... things

Answer 4

A garbage collector is never necessary, provided you have sufficient memory. However, in reality, we don't have infinite memory, and so we need some method to reclaim memory that is no longer needed. In impure languages like C, you can explicitly state you're done with some memory to free it - but this is a mutating operation (the memory you just freed is no longer safe to read), so you can't use this approach in a pure language. So it's either somehow statically analyze where you can free the memory (probably impossible in the general case), leak memory like a sieve (works great until you run out), or use a GC.

Answer 5

As others have already pointed out, Haskell requires automatic, dynamic memory management: automatic memory management is necessary because manual memory management is unsafe; dynamic memory management is necessary because for some programs, the lifetime of an object can only be determined at runtime.

For example, consider the following program:

main = loop (Just [1..1000]) where
  loop :: Maybe [Int] -> IO ()
  loop obj = do
    print obj
    resp <- getLine
    if resp == "clear"
     then loop Nothing
     else loop obj

In this program, the list [1..1000] must be kept in memory until the user types "clear"; so the lifetime of this must be determined dynamically, and this is why dynamic memory management is necessary.

So in this sense, automated dynamic memory allocation is necessary, and in practice this means: yes, Haskell requires a garbage collector, since garbage collection is the highest-performance automatic dynamic memory manager.

However...

Although a garbage collector is necessary, we might try to find some special cases where the compiler can use a cheaper memory management scheme than garbage collection. For instance, given

f :: Integer -> Integer
f x = let x2 = x*x in x2*x2

we might hope for the compiler to detect that x2 can safely be deallocated when f returns (rather than waiting for the garbage collector to deallocate x2). Essentially, we are asking that the compiler perform escape analysis to convert allocations in to garbage-collected heap to allocations on the stack wherever possible.

This is not too unreasonable to ask for: the jhc haskell compiler does this, although GHC does not. Simon Marlow says that GHC's generational garbage collector makes escape analysis mostly unnecessary.

jhc actually uses a sophisticated form of escape analysis known as region inference. Consider

f :: Integer -> (Integer, Integer)
f x = let x2 = x * x in (x2, x2+1)

g :: Integer -> Integer
g x = case f x of (y, z) -> y + z

In this case, a simplistic escape analysis would conclude that x2 escapes from f (because it is returned in the tuple), and hence x2 must be allocated on the garbage-collected heap. Region inference, on the other hand, is able to detect that x2 can be deallocated when g returns; the idea here is that x2 should be allocated in g's region rather than f's region.

Beyond Haskell

While region inference is helpful in certain cases as discussed above, it appears to be difficult to reconcile effectively with lazy evaluation (see Edward Kmett's and Simon Peyton Jones' comments). For instance, consider

f :: Integer -> Integer
f n = product [1..n]

One might be tempted to allocate the list [1..n] on the stack and deallocate it after f returns, but this would be catastrophic: it would change f from using O(1) memory (under garbage collection) to O(n) memory.

Extensive work was done in the 1990s and early 2000s on region inference for the strict functional language ML. Mads Tofte, Lars Birkedal, Martin Elsman, Niels Hallenberg have written a quite readable retrospective on their work on region inference, much of which they integrated into the MLKit compiler. They experimented with purely region-based memory management (i.e. no garbage collector) as well as hybrid region-based/garbage-collected memory management, and reported that their test programs ran "between 10 times faster and 4 times slower" than pure garbage-collected versions.

Answer 6

Haskell is a non-strict programming language, but most implementations use call-by-need (laziness) to implement non-strictness. In call-by-need you only evaluate stuff when it is reached during runtime using the machinery of "thunks" (expressions that wait to be evaluated and then overwrite themselves, staying visible for their value to be reused when needed).

So, if you implement your language lazily using thunks, you have deferred all reasoning about object lifetimes until the last moment, which is runtime. Since you now know nothing about lifetimes, the only thing you can reasonably do is garbage collect...