Create an insertion-ordered hash set using the provided
equals
and hashCode
.
The provided equals
must define a stable equivalence relation, and
hashCode
must be consistent with equals
. If the equals
or hashCode
methods won't work on all objects, but only on some instances of E, the
isValidKey
predicate can be used to restrict the keys that the functions
are applied to.
Any key for which isValidKey
returns false is automatically assumed
to not be in the set when asking contains
.
If equals
or hashCode
are omitted, the set uses
the elements' intrinsic Object.operator==
and Object.hashCode,
and isValidKey
is ignored since these operations are assumed
to work on all objects.
If you supply one of equals
and hashCode
,
you should generally also to supply the other.
If the supplied equals
or hashCode
functions won't work on all E
objects, and the map will be used in a setting where a non-E
object
is passed to, e.g., contains
, then the isValidKey
function should
also be supplied.
If isValidKey
is omitted, it defaults to testing if the object is an
E
instance. That means that:
new LinkedHashSet<int>(equals: (int e1, int e2) => (e1 - e2) % 5 == 0,
hashCode: (int e) => e % 5)
does not need an isValidKey
argument, because it defaults to only
accepting int
values which are accepted by both equals
and hashCode
.
If neither equals
, hashCode
, nor isValidKey
is provided,
the default isValidKey
instead accepts all values.
The default equality and hashcode operations are assumed to work on all
objects.
Likewise, if equals
is identical, hashCode
is identityHashCode
and isValidKey
is omitted, the resulting set is identity based,
and the isValidKey
defaults to accepting all keys.
Such a map can be created directly using LinkedHashSet.identity
.
Source
external factory LinkedHashSet({bool equals(E e1, E e2), int hashCode(E e), bool isValidKey(potentialKey)});