cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dennis Dam <d....@hippo.nl>
Subject cocoon 2.1.8 - RepeaterJXPathBinding, order of rows is inverted
Date Thu, 15 Jun 2006 13:02:08 GMT
Hello all,

I found a bug in the forms block of Cocoon 2.1.8 / 2.1.9. I found that 
putting a min-size / initial-size attribute on a repeater element in the 
CForms model, inverts the order of rows upon binding the form. So for 
example the input document looked like:

<document>
  <row>a</row>
  <row>b</row>
  <row>c</row>
  <row>d</row>
</document>

and my repeater model element's intial-size was set at 3. After 
transforming the forms template with the forms transformer, I see the 
following field instances (pseudocode):

<field>c</field>
<field>b</field>
<field>a</field>
<field>d</field>

So what happens? The JXPathbinding for the repeater inverts the order of 
the elements with index < initial-size! I traced this bug back to the 
org.apache.cocoon.forms.binding.RepeaterJXPathBinding class, where I 
found the following code snippet in the doLoad() method:

            while (rowPointers.hasNext()) {
                // create a new row, take that as the frmModelSubContext
                Repeater.RepeaterRow thisRow;
                if (initialSize > 0) {
                    thisRow = repeater.getRow(--initialSize);
                } else {
                    thisRow = repeater.addRow();
                }


I changed this into:

        int currentRow = 0;
        while (rowPointers.hasNext()) {
            // create a new row, take that as the frmModelSubContext
            Repeater.RepeaterRow thisRow;
            if (currentRow < initialSize) {
                thisRow = repeater.getRow(currentRow++);
            } else {
                thisRow = repeater.addRow();
            }

and now the binding works correctly / like I want it to work :). I must 
note that I do NOT use an identity in my repeater binding declaration. 
But I think it would be logical that leaving out the identity element 
should result in preservation of the order the elements occur in the 
input document.

Is there a reason for inverting the binding order of the "minimum set of 
rows"? If not, could someone fix this issue in 2.1.8 / 2.1.9?

regards,
Dennis Dam





Mime
View raw message