pivot-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "aappddeevv" <aappdde...@verizon.net>
Subject RE: [jira] Resolved: (PIVOT-542) enable the serializer to initialize components written in java code
Date Wed, 23 Jun 2010 17:58:56 GMT
A few thoughts:

The seriliazer already plays the initialization role.  You just mentioned a
bunch of alternatives that are simpler using this approach actually. Using
bxml:include and adding it to the host component (1 up) is a lot less
natural than "just use the class name like everywhere else" to use the
component.

My understanding is that skins are views. Components are model/controllers.
The view should not be loading user data and other child content into the
mode/controller so I'm not sure I would agree that the skin should be doing
this at all because part of the initialization is also Component data such
as userData.

With the editor, there are a few ways to do this. This is actually a very
direct approach. In the thoughts below, I am still forced, at a minimum, to
use an interface and for me to write code for wiring. I'm not against
writing code but when the serializer can do it naturally, that's a win.

This is not a big thought improvement, but incremental. It feels very
natural approach. I personally found the other approaches to be more
difficult (not impossible though).  It took me awhile to understand why I
had to always had to do extra work with includes or setting left and right
children in code until I realized that the serializer should be doing it for
me. The serializer needs a root, why not let the programmer insert the root
and have the serializer build around it.  After all that's what the
programmer is forced to do manually anyway.

I'm not sure that XAML compiles to classes actually. I think it just
compiles to binary XML then loads it in a partial class that you cannot see.
Then the WPF classes all call InitializeComponent() in their constructor. I
don't know the details of WPF though. I'll look at MXML, had not heard of
it.

Thanks for your thoughts.


-----Original Message-----
From: Greg Brown (JIRA) [mailto:jira@apache.org] 
Sent: Wednesday, June 23, 2010 1:06 PM
To: dev@pivot.apache.org
Subject: [jira] Resolved: (PIVOT-542) enable the serializer to initialize
components written in java code


     [
https://issues.apache.org/jira/browse/PIVOT-542?page=com.atlassian.jira.plug
in.system.issuetabpanels:all-tabpanel ]

Greg Brown resolved PIVOT-542.
------------------------------

    Resolution: Won't Fix

OK, thanks for the example. I hate to say it, but even after seeing how it
might be used in practice, I'm still not a fan of the approach. As I
mentioned in another thread, the recommended way to create a custom
component is to also create a custom skin that goes with it. So, I'd expect
to see an EditorSkin class that uses BXMLSerializer internally to load and
bind() to its content. Alternatively, you could have your Editor class
implement Bindable and use <bxml:include> to add it to the host component
(Bindable is effectively the equivalent of "code-behind" in Pivot).

If you don't want to create a custom skin and don't want to use an include,
you could still use BXMLSerializer in the Editor constructor to load the
left and right panes and set them via setLeft() and setRight(). After
loading each pane, you could call bind() on the editor instance (this) to
attach any values declared in the includes to the Editor instance.

While what you suggest makes sense for frameworks that compile XML to
classes (e.g. MXML, XAML), it just isn't consistent with the serializer
concept. Serializers take an input stream and return an object. The object
that creates the serializer isn't meant to also be the object that is
instantiated by the serializer. That's just a little too convoluted.

Thanks for the submission, though.


> enable the serializer to initialize components written in java code
> -------------------------------------------------------------------
>
>                 Key: PIVOT-542
>                 URL: https://issues.apache.org/jira/browse/PIVOT-542
>             Project: Pivot
>          Issue Type: Improvement
>            Reporter: Appddevvv
>         Attachments: initialize-examples.zip,
WTKXSerializer-with-initialize.patch
>
>
> The current serializer performs two funtions: creating objects and
initializing them. The typical use case is to create your component (say
java code) then call the serializer to create the content. Then within your
class, attach the content and initialize your object in various application
specific ways. However, since your object is typically a java object and
often a Component subclass, for those writing new components as composites
of others, you have to manually write some (not all) initialization code
that could also be performed in the BXML file. 
> To initialize your object using bxml content, you can use annotations etc.
However, if the bxml file also defines user data or other elements, you also
have to manually bring that in from the bxml and do some wiring. In
addition, it is currently impossible to initialize some pivot objects that
you subclass from, say the splitpane, with content from bxml without having
to create an intermediate container or use two bxml files for the left and
right child. For example, if you subclass from SplitPane, an underlying bxml
file that holds your content must have another container in order to define
your left and right children in or you have to use two separate bxml files
per child--this may or may not be a good thing and complicates the
implementation.
> To support the use case of creating subclasses of Components (and pivot
likes to use OO inheritance) and allowing the initialization of that
component to be matched by a bxml file, the serializer needs to be able to
"apply" itself to the component as if the bxml file and the subclassed
component were a reflection of the other. Bindable was designed to help with
this, however, the bxml file is still considered at level below (the
content) of the Component you are writing.
> The benefits are:
> a) A cleaner code style that has implied semantics--i.e. this bxml file is
the mirror image of your subclassed component at the same logical tree
level.
> b) More consistent coding conventions for novice programmers.
> The capability to do this can be implemented in different ways including
the current approaches of course. The small patch allows a new style of
Component creation blending code and bxml in a more seamless way. Its no
longer something you instantiate to "get" content, it is the content. It
helps Component writers who want to simply subclass another component. It
looks like a trivial change but it is a different perspective on
initialization that has natural semantics for programmers. It also reduces
complexity for creating subclasses of Components that have properties
representing the tree e.g. SplitPane.
> This is similar in spirit to "code-behind" in other frameworks.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message