[html-formfu] Patch for H::F::ObjectUtil::remove_element issue, submitted poorly :(

[Mike South]

On Sat, May 30, 2009 at 3:42 PM, Carl Franks <fireartist at gmail.com> wrote:
> 2009/5/30 Mike South <msouth at gmail.com>:
>>
>> I had a Repeatable block which had a delete_if_true checkbox in a
>> child block.  The remove_element call that gets rid of the
>> delete_if_true box on an empty_row was croaking (this is because
>> get_fields recurses, but remove_element doesn't).
>>
>> I submitted a patch with a test on google code, but I did the
>> summary/name poorly and I didn't identify it as a bug, etc.  (First
>> time using google code, I didn't realize I wouldn't be able to update
>> what I submitted).  Sorry about that!
>
> Hi Mike,
>
> Sorry, this is the intended behaviour, and I'm not keen on changing it.
> Though, you could use the following idiom to achieve what you want:
>
>    $field->parent->remove_element( $field );

I could, but it's not my code that's dying, it's HTML::FormFu::Model::DBIC.

The way it is now, if someone puts that checkbox in a sub block, say
for formatting (there's an example of someone that tried to do this on
the list earlier, I'm not the first that's tried), DBIC croaks on it.

In other words, applying the following patch causes the test to fail:

Index: t/default_values/has_many_repeatable_delete_true.yml
===================================================================
--- t/default_values/has_many_repeatable_delete_true.yml	(revision 1525)
+++ t/default_values/has_many_repeatable_delete_true.yml	(working copy)
@@ -21,10 +21,12 @@
       - type: Text
         name: address

-      - type: Checkbox
-        name: delete
-        model_config:
-          delete_if_true: 1
+      - type: Block
+        elements:
+          - type: Checkbox
+            name: delete
+            model_config:
+              delete_if_true: 1

   - type: Hidden
     name: count

If you intend to disallow that, I think it ought to be pointed out in
the documentation that you need to keep the Repeatable block flat.

If you want to keep non-recursion as the default behavior in
ObjectUtil, a couple of alternatives would be:

(1) Alter H::F::M::DBIC to eval the remove_element and recurse into
child blocks to handle a case like this without croaking

(2) Add a 'recurse' flag to remove_element, and use that in
H::F::M::DBIC's call (or even a remove_element_recursively method).

I would be willing to code either of those up, just let me know.

mike

>
> Carl
>
> _______________________________________________
> HTML-FormFu mailing list
> HTML-FormFu at lists.scsys.co.uk
> http://lists.scsys.co.uk/cgi-bin/mailman/listinfo/html-formfu
>