EnumSet, what the ?

[Roedy Green]

I discovered that only the static members of EnumSet are public.

In other words, EnumSets appear to be immutable at least to anyone
but Sun.

This is highly puzzling. Why bother with a bitset representation for a
set without any mathematical set operations? There is complement, but
where are union (or), intersection (and), set difference (xor),
bitwise operations? Equals at first looks to be but they defined it
as object identity not bit map identity.

The general purpose "of" method works mostly with enum constant
operands, not with entire EnumSets.

Perhaps I should simply rewrite EnumSet.


--
Bush crime family lost/embezzled $3 trillion from Pentagon.
Complicit Bush-friendly media keeps mum. Rumsfeld confesses on video.
http://www.infowars.com/articles/us/mckinney_grills_rumsfeld.htm

Canadian Mind Products, Roedy Green.
See http://mindprod.com/iraq.html photos of Bush's war crimes

 

[George Cherry]

I discovered that only the static members of EnumSet are public.

In other words, EnumSets appear to be immutable at least to anyone
but Sun.

This is highly puzzling. Why bother with a bitset representation for a
set without any mathematical set operations? There is complement, but
where are union (or), intersection (and), set difference (xor),
bitwise operations? Equals at first looks to be but they defined it
as object identity not bit map identity.

Have you looked at EnumSets' super classes, AbstractSet and
AbstractCollection? Maybe their bulk operations will give you
what you want. For example, AbstractSet overrides the equals
in Object so that it gives you true set equality, not object equality.

George W. Cherry

 

[Roedy Green]

Perhaps I should simply rewrite EnumSet.

I discovered another peculiar feature of RegularEnumSet

look at this code:

void addRange(E from, E to) {
elements = (-1L >>> (from.ordinal() - to.ordinal() - 1)) <<
from.ordinal();
}

He is constructing a bit mask of ones for the range, 1=included,
0=excluded, 0=unused, counting bits from the lsb. e.g. binary
000111111100 for ordinals 2 through 8. Then instead of doing elements
|= for addRange he does elements = which really should have been
called setRange.

Growl!



--
Bush crime family lost/embezzled $3 trillion from Pentagon.
Complicit Bush-friendly media keeps mum. Rumsfeld confesses on video.
http://www.infowars.com/articles/us/mckinney_grills_rumsfeld.htm

Canadian Mind Products, Roedy Green.
See http://mindprod.com/iraq.html photos of Bush's war crimes

 

[Jesper Nordenberg]

I discovered that only the static members of EnumSet are public.

In other words, EnumSets appear to be immutable at least to anyone
but Sun.

EnumSet is not immutable, it implements the mutating methods in the
Set interface.

This is highly puzzling. Why bother with a bitset representation for a
set without any mathematical set operations? There is complement, but
where are union (or), intersection (and), set difference (xor),
bitwise operations? Equals at first looks to be but they defined it
as object identity not bit map identity.

An EnumSet is not a bitset, so it shouldn't contain any bit operations
and correctly doesn't. Sun has chosen to internally represent the set
as a bitset, but that's another issue. The EnumSet implements all
methods in the Set interface, so union == addAll(), intersection ==
retainAll(), difference == a combination of retainAll(), addAll() and
removeAll().

/JN

 

[George Cherry]

EnumSet is not immutable, it implements the mutating methods in the
Set interface.


An EnumSet is not a bitset, so it shouldn't contain any bit operations
and correctly doesn't. Sun has chosen to internally represent the set
as a bitset, but that's another issue. The EnumSet implements all
methods in the Set interface, so union == addAll(), intersection ==
retainAll(), difference == a combination of retainAll(), addAll() and
removeAll().

Yes, I said something similar (but not so well)
in an earlier post to Roedy. Hey Roedy, Java's
a high-level language; get your mind out of the
bit gutter. : o )

George

 

[Roedy Green]

Yes, I said something similar (but not so well)
in an earlier post to Roedy. Hey Roedy, Java's
a high-level language; get your mind out of the
bit gutter. : o )

my background is math. I expect operations like union, intersection,
on something that calls itself a set, which map very nicely to what an
assembler programmer like me does with bitmaps. It turns out the
interesting methods of an EnumSet are way down in the AbstractSet
class.

There seems to be a heck of a lot of dithering compared with the way I
implemented such a feature in Abundance.

--
Bush crime family lost/embezzled $3 trillion from Pentagon.
Complicit Bush-friendly media keeps mum. Rumsfeld confesses on video.
http://www.infowars.com/articles/us/mckinney_grills_rumsfeld.htm

Canadian Mind Products, Roedy Green.
See http://mindprod.com/iraq.html photos of Bush's war crimes

 

[George Cherry]

my background is math. I expect operations like union, intersection,
on something that calls itself a set, which map very nicely to what an
assembler programmer like me does with bitmaps. It turns out the
interesting methods of an EnumSet are way down in the AbstractSet
class.

Okay, but Joshua Bloch's elegant Collections Framework
gives you all the set operations--and they work between different
Set implementations (including EnumSet) very conveniently.
EnumSet implements the Collections Framework's root
interface, Collection, and, of course, its subinterface, Set.
It turns out that EnumSet's relevant bulk operations are
way up in the Collection interface, and they are defined
explicitly as set operations in the API for the Set interface.
(Are interfaces neat or what?)

Bloch's excellent tutorial on the Java Collections Framework
is at

http://java.sun.com/docs/books/tutorial/collections/

The following is from the tutorial

The bulk operations are particularly well suited to Sets; when applied to
sets, they perform standard set-algebraic operations. Suppose s1 and s2 are
Sets. Here's what the bulk operations do:
a.. s1.containsAll(s2): Returns true if s2 is a subset of s1. (s2 is a
subset of s1 if set s1 contains all the elements in s2.)
b.. s1.addAll(s2): Transforms s1 into the union of s1 and s2. (The union
of two sets is the set containing all the elements contained in either set.)
c.. s1.retainAll(s2): Transforms s1 into the intersection of s1 and s2.
(The intersection of two sets is the set containing only the elements that
are common to both sets.)
d.. s1.removeAll(s2): Transforms s1 into the (asymmetric) set difference
of s1 and s2. (For example, the set difference of s1 - s2 is the set
containing all the elements found in s1 but not in s2.)

There seems to be a heck of a lot of dithering compared with the way I
implemented such a feature in Abundance.

Fooling around directly with bits is a heck of a lot of dithering compared
with the way Bloch specified and implemented set operations in the
Java Collections Framework--and you still get the efficiency of bit
vector dithering, because the implementation of EnumSet uses bit
vectors behind the high level scene.

George W. Cherry