Kedar,
Yes, I think that's the point: a missing attribute is quite
different from an empty one--and should be allowed.
Lloyd
On Aug 8, 2006, at 5:35 PM, Kedar Mhaswade wrote:
>> Backend treats "" empty String as removal of the property itself.
>> This mean, unless backend changed the behaviour or have a way for
>> setting property value to "" empty string, there is not much GUI
>> can do.
>> I am transferring this bug to 'admin'. Once they design and
>> implemented a way to set empty string value to Property, then GUI
>> can implement that.
>> Here is comment from bug# 6363330
>>> Set empty valued property is treated by backend as request to
>>> remove property.
>>> This is only way to remove property from CLI, as far as it has
>>> only 'get' and 'set' command to handle dotted named attributes
>>> and properties.
>>> We can not re-consider this behaviour now because of backward
>>> compartibility.
>>> Kedar and Abhijit proposed to have the special value for property
>>> to set it to "empty" string.
>>>
>
> The point that I was trying to make was it is hard to represent on any
> interface. Definitely not on the lines of "It is someone else's
> problem" or
> "something is not done right".
>
> What is true for properties is however not true for attributes, that's
> all. For example, if I am modifying an attribute (let's say per the
> schema,
> this is an IMPLIED attribute) of an element on the GUI, then when I
> specify an empty value in the text field, that attribute is removed
> from
> domain.xml.
>
> How would we (easily) make this to be distinct from having the
> attribute with
> a value "" in the domain.xml because
>
> <foo a1="" a2="xyz" .../>
>
> could be a configuration different from:
>
> <foo a2="xyz" .../>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe_at_glassfish.dev.java.net
> For additional commands, e-mail: dev-help_at_glassfish.dev.java.net
>