Re: Vector versus Hashtable and Related Questions
Russ wrote:
The initial question is: Is a Vector or Hashtable efficient for
I wouldn't use either one. Those classes were supplanted in 1998 by the
various implementations of 'List' and 'Map', specifically 'ArrayList' and
'HashMap'. Don't use 'Vector' or 'Hashtable' unless forced to.
searching for clients in say a server to deliver a message? Looking at
this link:
http://www.devx.com/tips/Tip/25941
It says: "Searching through a Vector is O(n), which means that given n
elements, you may have to look at every element before finding what
you are looking for. Unless you use a value as its own key, a
Hashtable will be equally poor."
That's nearly correct. A value need not be its own key.
However that "Unless" creates questions, because looking at some old-
school code (Java 1) from an open-source server where the coder (who
has far more skill than myself) wrote the following method where
"users" is a Vector:
synchronized VUser findUser(String name) {
for (int i = 0; i< users.size(); i++) {
Do not embed TAB characters in Usenet code posts!
VUser user = (VUser) users.elementAt(i);
if (user.name.equals(name)) {
return user;
}
}
return null;
}
Why did the coder do it that way? Isn't "users" as a Hashtable more
efficient?
Depends on the context that you didn't show.
BTW, that code would look different nowadays, assuming the same O(n) algorithm:
synchronized VUser findUser(String name) {
for ( VUser user : users ) {
if (user.name.equals(name)) {
return user;
}
}
}
synchronized VUser findUser(String name) {
VUser user = (VUser) users.get( name );
return user;
}
Why would you bother with a cover method for a simple 'get()' call?
It seems obvious that if "users" is a Hastable and once the instance
of VUser is created to simply map the Value to its own key of a
protected String name within the Value:
VUser VU = new vUser( String name );
users.put( VU.name, VU );
Also, why use "synchronized" on the findUser method when both Vectors
and Hashtables are already synchronized?
Ii assume you're talking about the O(n) search. Only individual methods are
synchronized. The search involves multiple accesses across multiple method
calls, requiring additional synchronization.
Finally, suppose I create an instance of a class and put that instance
in two different Hashtables
or Vectors. If I change a variable in that instance in one Hashtable
won't the same instance in the other Hashtable be changed or updated
as well? Common sense says it is the same allotment of memory so yes,
but I am just trying to be certain.
Variables, or references, can point to the same object, so yes.
Again, use neither Hashtable nor Vector, and do use generics.
--
Lew