The type of keys on which resources in the cache are indexed.

The type of caches holding bindings from key to 'resource

create destroy n creates a cache with a size-bound of n. Remember that the size-bound is not upheld strictly by all caches. Moreover, caches instantiated with a specialised size (i.e., empty and singleton caches) ignore the size parameter entirely.

put c k v binds the key k to the resource r in the cache c and transfer the ownership of r to the cache. After put, it is the cache's responsibility (not yours) to clean-up the resource.

If k is already bound to a value in c, the previous binding disappears and is replaced by the new binding to r. If this happens, the resource for the previous binding is cleaned-up by the cache.

If the cache is already at capacity, put may cause another binding to be removed from the cache to make room for the new one. If this happens, the resource of the removed binding is cleaned-up by the cache.

take c k removes the binding of k from c and returns it to the caller along with the ownership of the associated resource: you are now responsible for cleaning-up the resource.

If k is not bound in c, then take c k does nothing.

take_all c returns the list of bindings held by the cache along with the ownership of their resources (you are now responsible for cleaning-up) and clears the cache entirely.

take_some c f returns a list of bindings (k,r) such that f k r is true. The returned bindings are removed from the cache and the ownership of their resources is transfered to the caller (you).

borrow c k f calls f with r if k is bound to r in c. This does not remove the resource from the cache: the cache is still responsible for cleaning-up the resource.

It is unsafe to clean-up the borrowed resource from within f.

It is unsafe to use the cache c from within the function f. If you need to do so, you can adopt either of these two approaches:

• use take, use the resource, and use put,
• make f returns a list of operations to perform on the cache and process this after f has returned.

Note that the in caches with a non-FIFO replacement policy, this may have a side effect on the k-to-r binding. Specifically, in those caches, it might make it less likely to be removed when supernumerary bindings are inserted.

fold f c init folds the function f and value init over the bindings of c from newest to oldest.

At each called to f, the resource of the traversed binding is borrowed by f. Consequently, the same limitations apply for fold as for borrow.

It is unsafe to clean-up any of the borrowed resources.

It is unsafe to use the cache from within f. If you need to do so, you can adopt either of these two approaches:

• use take_all, fold over the list of bindings, use put on each binding,
• maintain a list of operations to perform on the cache within the fold accumulator and process this list once the folding is over.

fold_oldest_first is like fold but in reversed order: the elements that would be the first to be removed are traversed first. In a FIFO cache, it is oldest-first traversal.

The same limitations and warning applies as for fold.

remove c k removes and cleans-up the binding from k in c. If k is not bound in c, it does nothing.

clear c removes and cleans-up all bindings from c.

filter c f removes and cleans-up all the bindings (k, v) such that f k v = false.

length c is the number of bindings held by c.

capacity c is the number of bindings c can hold: capacity (create n) = n