I'll just add: corelibs-dev is the right place for this discussion, not
amber-dev.
On 10/16/2025 11:05 PM, Ethan McCue wrote:
I don't have anything intelligent to add, but I assume either
* It's a good / okay idea, not high priority
* putIfAbsent and computeIfAbsent are seen to be enough for lazy
operations
* Something much more subtle and/or annoying about how it will affect
the universe of map implementations.
Either way, just so there is something concrete to talk about:
import module java.base;
public final class Maps {
private Maps() {
}
private static final Object DEFAULT = new Object();
@SuppressWarnings({"unchecked","rawtypes"})
public static <K, V> V getOrCompute(
Map<K, ? extends V> m,
Object key,
Function<? super K, ? extends V> compute
) {
Object value = ((Map) m).getOrDefault(key, DEFAULT);
if (value == DEFAULT) {
return compute.apply((K) key);
}
else {
return (V) value;
}
}
}
(Note the three needed casts - this might be difficult for the same
reason Map#gettakes an Objectand not a K)
On Thu, Oct 16, 2025 at 7:20 PM Chris Bouchard
<[email protected]> wrote:
Alberto,
I believe your getOrPut methods already exist as putIfAbsent and
computeIfAbsent, unless I'm missing a subtle difference.
On Thu, Oct 16, 2025, 14:20 Alberto Otero Rodríguez
<[email protected]> wrote:
Also, other two new methods might be interesting if you want
to get a value from a map, but if the key doesn't exist you
want to insert that value in the map and return it:
default V getOrPut(Object key, V defaultValue)
default V getOrPut(Object key, Function<? super K, ? extends
V> defaultValueFunction)
Chris
