On Tue, 27 Jul 2010, Greg Brown wrote:
You want me to create a one line file and include it because Pivot
doesn't support lifecycle notification? That's absurd. A one line
bxml file. Is there any conceivable use for a one line bxml file aside
from working around this problem? And I'm supposed to create these
files every time I want a notification?
The point is that BXML (intentionally) doesn't define any kind of
"lifecycle notification". It is a shortcut for creating Java class
instances. Java's support for "lifecycle notifications" includes the
constructor, methods, and the finalizer. You have access to the same
facilities in BXML.
The Bindable#initialize() method only exists because, without it, a root
element would have no way of knowing that all of the page variables had
been fully processed. In essence, initialize() is a "constructor" for
functional modules defined in BXML. Just like you partition
functionality into classes in Java, you can partition it into separate
includes in BXML.
Here's the thing. I have a component. It has a button for adding
children, but it has to know what wtkx file to include and it has to
know the value of one other property. I want it to create one child
(like pressing the 'add child' button) after its properties have been
set (so it knows how to do it) by the serializer so that one child item
appears in the UI when the user sees it. From there they can click the
'add child' button to their heart's content.
This makes an even stronger case for separating your BXML file into
includes. In the BXML that declares your button, you can include the
BXML that defines the child component. Then, the button press listener
can use BXMLSerializer to load new instances of the child component
every time it is pressed.
Now, with that said, I would like to provide some hopefully constructive
feedback. While I sincerely appreciate your interest in the platform and
suggestions for how to improve it, your approach to communication on the
mailing lists often borders on antagonistic and arrogant when those
suggestions are rejected. This makes me (and possibly others) reluctant
to want to help you.
I recommend that you try asking questions about how your design issues
might be solved within the existing framework before suggesting changes
to it. Note that I'm not attempting to discourage you or anyone else
from challenging elements of Pivot's design - that's how the platform
evolves. However, a lot of thought has gone into the existing design,
and I think it's pretty good. The constant onslaught is getting a bit
frustrating.
I just read this message and I knew I was going to get it. ;)
I'm sorry. I let my frustration get the better of me and I came out
swinging. For that, I apologize. I really appreciate the dialog and
interaction I have with you and other Pivot developers and users on this
mailing list, and I don't intend to abuse that privilege.
As regards this particular issue, I'm still using a serialization callback
to accomplish some things--including a new use case---but I'm going to
give it a rest for now. I'm not ready to discuss my new use case. It may
be solved by pivot 2, which I haven't migrated to yet.
Cheers,
Michael
