Agreed. The port to Trinidad did _not_ cause this problem. My initial diagnosis was incorrect and your point on Tiles demonstrating the problem is the correct diagnosis.

Are you in a position to say if this fix will make it into 1.2.2?

                                        -=> Gregg <=-

Adam Winer wrote:
Oh, absolutely, it's a bug, and we'll apply the fix you've provided.
I'm just pointing out that it's a really long-standing one, and that:

'This worked fine in ADF Faces before
porting to Trinidad where the listName was just "shuttle1:leading".
Note, however, that Trinidad now adds an additional "content" prefix'

... isn't quite right, since the bug existed in ADF Faces too,
and Trinidad doesn't add the "content" prefix itself.

-- Adam


On 9/4/07, Gregg Leichtman <[EMAIL PROTECTED]> wrote:
Having thought about this following your FYI (thank you for the hint), I
realized that Tiles must indeed be adding a tiles name prefix to the
name and ID attributes. The shuttle component is in a tile whose
definition name is, not surprisingly, "content".

Given this, however, I still would classify this behavior in Trinidad as
a bug based on the following reasoning. The component code that is
generating the _javascript_ variables:

        content_shuttle1_leading_desc
        content_shuttle1_trailing_desc

appears to use a global replace of underscore for colon (I didn't check
the Java code for this, but it seems reasonable based upon the generated
ID); however, the code in the generated _javascript_ function
TrShuttleProxy._getDescArray that searches for an element match via ID
uses a single or first replace of underscore for colon. This
inconsistency in the handling of the modification of element IDs will
always cause a problem for any code, Tiles or otherwise that adds
something with a colon to the ID.

Therefore, I would re-categorize this as an integration bug instead of a
general Trinidad bug, but I would suggest that it is still a bug in
Trinidad that should be fixed. If nothing else, I will need to,
hopefully temporarily, create a duplicate TrShuttleProxy._getDescArray
function with the suggested patch in order to get this functionality to
work with Tiles or alternatively modify the Java generation code,
something I would rather not do, since the latter interferes with easy
upgrades.

Do you agree that this is a bug and that it should be fixed?

                                                              -=> Gregg <=-

Adam Winer wrote:
> Gregg,
>
> FYI, I don't think Trinidad is adding the "content:".   Perhaps
> Tiles 2.0 is (I've not looked at the newer version).
>
> -- Adam
>
>
> On 9/3/07, Gregg Leichtman <[EMAIL PROTECTED]> wrote:
>
>>  I have found a bug in the generated _javascript_ for the Select*Shuttle
>> components. A patch follows:
>>
>>  TrShuttleProxy._getDescArray = function(listName) {
>>  //  var descArray = window[listName.replace(':','_') + '_desc'];
>>    var descArray = window[listName.replace (/:/g,'_') + '_desc']; // gsl fix
>>    return descArray;
>>  }
>>
>>  where listName is content:shuttle1:leading.
>>
>>  The problem is that the listName is not using a global replace; therefore
>> only the first ':' is being replaced. This worked fine in ADF Faces before
>> porting to Trinidad where the listName was just "shuttle1:leading". Note,
>> however, that Trinidad now adds an additional "content" prefix to the
>> generated inline _javascript_ variables:
>>
>>  content_shuttle1_leading_desc=new Array('The First Item
>> Desc',
>>  'The Second Item Desc',
>>  'The Third Item Desc','');
>>  content_shuttle1_trailing_desc=new Array('');
>>
>>  This means that a global replace needs to be done to create the proper
>> variable name for matching in the _javascript_ code. I will report this on
>> Jira for the Trinidad project.
>>
>>  I don't have access to the latest snapshots. The last one I see is from
>> July 11th at:
>>
>>
>> http://people.apache.org/repo/m2-snapshot-repository/org/apache/myfaces/trinidad/trinidad-impl/1.2.1-SNAPSHOT/
>>
>>  If this has been fixed already, then please clear my bug report which
>> should have the side benefit of letting people know that the bug has been
>> found and fixed already.
>>
>>                                     -=> Gregg <=-
>>



Attachment: signature.asc
Description: OpenPGP digital signature

Reply via email to