Enhanced SwingBuilder to add a columnModel into a table, added cellEditor support to tables and trees, implemented optional DefaultRenderer-behaviour for cellRenderers

Added code tags.

Hamlet, since you are the expert here, what do you think of this?

The request and sample code seems reasonable. I will take a look at the implementation and commit tonight if it looks good. Also, I'll check with Andres to see if Griffon breaks.

The patch looks good however I have a few comments.

ColumnModelFactory

• checking if the current node is a JTable during newInstance() is a bit aggressive, as you can't create column models on separate closures (think a reusable column model making closure). I'd prefer the check to be added to the onNodeCompleted() method (which the factory already defines). Avoid throwing an exception, you can switch to logging a warning for example.

ClosureCellEditor

• it looks to me there's a chance that the callbacks field might throw an NPE if the 1st constructor is not used. I'd prefer to initialize the field to an empty Map, then add the provided callbacks (if any) using putAll() .
• why does ClosureCellEditor needs to implement GroovyInterceptable?

ColumnFactory

• same check for the current parent as with ColumnModelFactory, better move them to setParent()/onNodeCompleted where it fits.
• once you know the width attribute is a Number you can call .integerValue() on it, faster than 'as Integer'.

GetCellEditorFactory

• I'd prefer the node name to be just 'editorValue' not 'getEditorValue'

CellEditorFactory, CellEditorGetValueFactory, CellEditorPrepareFactory

• these factories use the builder's binding to store temporal data, it's better to use the builder's context. Each node has its own context. The current context points to its parent context (if any) so you can share data between children and parents while the children are being processed. Using the context is much better as builder variables might break using Griffon's CompositeBuilder.

CellEditorGetValueFactory, CellEditorPrepareFactory

• these factories assume an specific parent which is why they return the builder.current. Be aware that this might be a bug as the parent might be added to the grandparent. If you do not care about the current node then return an empty map (use Collections.emptyMap() to guard against undesired attributes).

Cheers
Andres

Alright, I will merge this patch in this weekend when I have an hour of free time to account for Andres' feedback. Great feedback Andres!

I applied all the comments from Andres.
I addition I added the last two comments to RendererFactory / RendererUpdateFactory aswell,
because this Factorys had been implemented the same way.

Alexander,

I am ready to commit this work, thank you.

However, I cannot get a working example running. Can you please update your code example in the JIRA description so that it runs in GroovyConsole?
The example from the unit test runs but is not a good test because it doesn't show any data or header info, as the sample here does.

THanks,

Hamlet,

here are two examples, runnable in a 1.7.6 GroovyConsole, one for a JTable, one for a JTree

JTable-Example

import groovy.swing.*
import javax.swing.*

new SwingBuilder().edt {
  frame(id: 'test', title: 'Frame', size: [300, 300], show: true) {
    borderLayout()
    def data = [ ['Chang', 'Peng', 30], ['John', 'Doe', 17], ['Hans', 'Meyer', 25]] as Object[][]
    scrollPane {
      table(new JTable(data, ['name', 'prename', 'age'] as Object[])) {
        columnModel {
          column('name', modelIndex: 0, headerValue: 'Lastname', width: [100, 150, 250]) {
            headerRenderer {
              onRender { children[0].text = value.toUpperCase() }
            }
          }
          column('prename', modelIndex: 1, headerValue: 'Prename', width: [100, 150]) {
            cellRenderer {
              label(foreground: java.awt.Color.RED)
              onRender { children[0].text = value }
            }
          }
          column('age', modelIndex: 2, headerValue: 'Age', width: 70) {
            cellEditor {
              prepareEditor { children[0].text = value*2 }
              editorValue { children[0].text.toInteger() }
            }
          }
        }
      }
    }
  }
}

JTree-Example

import groovy.swing.*
import javax.swing.*

new SwingBuilder().edt {
  frame(id: 'test', title: 'Frame', size: [300, 300], show: true) {
    borderLayout()
    scrollPane {
        tree(rootVisible: false, editable: true) {
          cellRenderer {
            onRender {
              children[0].text = value.userObject.toUpperCase()
            }
          }
          cellEditor {
            textField()
            prepareEditor {
                children[0].text = value.userObject
            }
            editorValue {
                value.userObject = children[0].text
            }
          }
        }
      }
   }  
}

fixed in head, rolling back to 1.7.6

rolled back to branch