If you are interested in data-centric, web-enabled software built in Java, my current initiative could interest you.

The move seems to be successful for now, however there is no anonymous access to the subversion repository yet. Will work on this soon.

What is webby’s status now? As a PoC, it is finished: It works, is generic, and could be used for other applications. However for general-purpose usage the design is not ideal: If the business model uses inheritance, webby will probably not have the right structure here and there (e.g. have in a related list of objects not just generic attributes, but also attributes of specialized classes); If you want to use “just” the bean editor in an existing application, this would not be too nice, because you have do specify one global “configurator” … and maybe in the application you’d have the need for two configurators; and you could ask the question, why a configurator is needed at all; also the OO design is not the nicest, it’s more a grown interface / class structure, not a nicely designed one.

I could imagine that with the lessons learnt and the PoC available, webby could be the starting point for a reusable artefact. The artefact I come to think about is a a component like a bean editor, whereas the bean editor is customizable to the needs of the end user. Defining in detail what requirements such an artefact could implement is outside of the scope of this posting, but stay tuned if you are interested.

1. Now you can sort and filter in lists, given that the dataprovider supports it. The new Webby4DBDataProvider and the Webby4DBNRelationDataProvider in the webby4db package do this nicely.
2. Easy control is available about which attributes and relations are visualized by explicitly creating the descriptor you want - don’t have to lean on the smartness of webby’s introspection: Even though introspection can be nice for creating a draft application quickly, sometimes you want to be more in control: Here this means controling what you see - not how you see it. We’ll come to this issue later…

Check out webby and the new Buco example!

With the current status of webby’s components you can see two things:

• A lot can be deduced from the business classes, namely attributes, their types, and relations between the business classes. The descriptor concept and the implementation of it (in the “Default…” and “Simple…” classes) hopefully prove this. The descriptor concept therefore seems to be a suitable measure towards webby’s goal, which is allowing me (and probably other developers) concentrating on business issues and algorithms, not repetitive coding of input/output/persistence stuff.
• Using the business objects’ structure (via descriptors) to allow visualization, editing, sorting, filtering, etc. is nice, but it makes the UI full, maybe overloaded, at least in some cases. Example: If you’ve got a relation from a contact person to all it’s email adresses (maybe 3 or 5, but never hundreds), filtering in a list of a person’s email adresses is quite useless.

The second point shall now be the trigger for the next step forward in bringing webby closer to it’s goals (
see “Why Webby?”):

In order to make the generic support for beans and relations work comfortably for the application’s user, the user needs sufficient control: This means for example control over sorting and filtering of lists (we have that now) and selecting what he wants to see (well, we don’t really have this, the developer select what the user sees, webby decides how to show it). Maybe the user would even like to decide how he wants to see attribute values or relations (e.g. by colors or icons).

The next (big) step of webby shall be into this direction, probably first by adding state to webby’s UI components, then by improving the components themselves.

Stay tuned!

The goal of this release is to have some small enhancements, whereas the most notable is probably the improved support for typical types of attributes: Whenever the SimpleAttributeDescriptor needs a Panel for an attribute, it uses UtilAttributes’s method guessAttributePanelClass. This method does a few checks via isAssignableFrom trying to find a better panel than the default TextFieldAttributePanel.

You can see the effect of the guessing in the examples, best in the Projects Example (click on a project’s name to see a list of the news): The boolean value News.published gets the BooleanAttributePanel; even nicer is the DateAttributePanel for the News.timestamp attribute: It uses wicket’s date/time selector!

Like it? Not? Let me know via a comment, if you like…

After announcing the availability of webby, my next goal is to determine where to go from here. Although I’ve got my own ideas (which can be partly read here), I am keen on getting feedback. I expect that some members of the wicket community will have an opinion about webby, some might even be interested in using it (if developed further into their preferred direction).

Please leave your feedback as comments to this post.

I am happy to receive any kind of feedback on it. In fact, the sole purpose (for the moment) of releasing it is to get feedback! So please don’t hold back and shoot on it and my thoughts!

One of the first things I might be doing is to explain what r8fe is actually about. Unfortunately this is (yet) hard to tell: Basically it is my playground, where I experimentally use some technology to work on my ideas about how to use this technology. Ugh, this is not really specific, is it? Well, wait for a future post to get more precise information.