+1

- robert

On Saturday, October 4, 2003, at 05:02 PM, Henri Yandell wrote:


I'm -1 to Lang. Comparators are not within its scope.


It does sound like the generic comparator is really just a factory method
for creating a structure of Collections comparators with a Bean comparator
structure. I believe it should be a static method on the BeanUtils
BeanComparator class. BeanUtils has a dependency on Collections if I'm not
mistaken, so can happily get the Collections reuse, without coding the
bean handling stuff again.

Hen

On Fri, 3 Oct 2003, Dirk Verbeeck wrote:

The easiest way to contribute to to integrate/add your code to an
existing project.
Commons-lang looks likes a good place to start:
http://jakarta.apache.org/commons/lang/api/org/apache/commons/lang/builder/
package-summary.html
http://jakarta.apache.org/commons/lang/api/org/apache/commons/lang/builder/
CompareToBuilder.html

But there is also:
http://jakarta.apache.org/commons/collections/api/org/apache/commons/collections/
ComparatorUtils.html
http://jakarta.apache.org/commons/collections/api/org/apache/commons/collections/
comparators/package-summary.html

Your "|" operator is a ComparatorChain
<http://jakarta.apache.org/commons/collections/api/org/apache/commons/collections/
comparators/ComparatorChain.html>

order=desc is a ReverseComparator <http://jakarta.apache.org/commons/collections/api/org/apache/commons/collections/
comparators/ReverseComparator.html>


and nulls=last can be done using NullComparator
<http://jakarta.apache.org/commons/collections/api/org/apache/commons/collections/
comparators/NullComparator.html>
and your getCustomer.getLastName lookup is something for beanutils, jexl
or jxpath

Your simple expression would be indeed a time saver for constructing a
custom comparator.
Did you already use some of the existing commons code?
If you did then I think either commons-lang or commons-collections will
be interested.

Dirk


Wickus Martin wrote:


A while back I wrote a generic comparator for Java objects and called it
GComp. Weve been using it internally at the company where I work, but I was
wondering if it might not be of value to the commons project.


Typically Java programmers create a custom Comparator implementation for
every comparison scenario they encounter. GComp on the other hand can
compare objects according to complex criteria based on a simple expression
that explains how the objects should be compared. Great time saver and
prevents problem of many small comparator classes bloating the code base.


Usage scenario:

Lets say we have a class called Customer with methods to access the first
name and last name. Lets also say we have a class called Order that holds a
reference to an instance of Customer.


To compare orders according to the customer surname (ascending) we could
create the following comparator:

new GComp("getLastName");

To compare a collection of orders according to the customer last name
(ascending) and then first name (descending) for those customers with the
same last name we could create the following comparator:


new
GComp("[EMAIL PROTECTED]|[EMAIL PROTECTED]
desc
");

To order null values last and ignore case we could create:

new
GComp("[EMAIL PROTECTED],nulls=last|getCustomer.
[EMAIL PROTECTED],nulls=last");


The above code only orders on string objects, but all objects implementing
the comparable interface is supported as is primitive types.


Currently GComp uses commons-logging to output debugging information if
commons logging is found on the classpath.

I'm happy to donate the code for further study. Let me know.

Keep well,
Wickus Martin






---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



Reply via email to