cxf-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Daniel Kulp <dk...@apache.org>
Subject Re: HashHappy
Date Tue, 29 Apr 2008 14:38:11 GMT
On Tuesday 29 April 2008, Fred Dushin wrote:
> Is there a reason hashed collections are used, almost exclusively, in
> CXF?  Even in cases where collections are predictably small,
> Hash(Maps| Sets) almost always win out over their sortable cousins in
> the java.util namespace, and this, even when the keys are sortable.
>
> Is there a technical reason for this?  A personal preference?

Depends on the location in the code.   One reason for using a hash based 
map is that there exists the ConcurrentHashMap that is a drop in 
replacement when thread concurrency is required.  Likewise, there is a 
CopyOnWriteArrayList for the ArrayList flip.   There aren't any 
conncurrent versions of the SortedSet/Map things.  (they were added in 
Java 1.6, but not available in Java 1.5) 


> I've done some performance comparisons between the 2, and for small
> collections, the log(n) lookup doesn't make a difference at all.  So
> for small collections, I'd think CXF might benefit from the relatively
> smaller memory footprint of sortable collections.  We'd also benefit
> from more predictable behavior, as ordering is deterministic, with
> sortable collections.
>
> The same goes for ArrayLists vs LinkedLists, BTW.

Actually, that really does depend.   In most cases, and ArrayList uses 
less memory than a LinkedList, especially if we set a "smart" initial 
size. 

> Just wondering...

Probably just preference and maybe a small bit of "consistency" thrown in 
for good measure.  


-- 
J. Daniel Kulp
Principal Engineer, IONA
dkulp@apache.org
http://www.dankulp.com/blog

Mime
View raw message